1 00:00:00,000 --> 00:00:07,660 Welcome to Cannabis Data Science. 2 00:00:07,660 --> 00:00:10,560 Thanks for holding tight. 3 00:00:10,560 --> 00:00:12,660 Promise it's going to be worth your while today. 4 00:00:12,660 --> 00:00:17,680 So have yet another new idea to share with you today. 5 00:00:17,680 --> 00:00:21,880 And also want to share a lot of ideas about the group and kind of get everyone on the 6 00:00:21,880 --> 00:00:22,880 same page. 7 00:00:22,880 --> 00:00:25,720 Kind of talk about what we're doing here. 8 00:00:25,720 --> 00:00:29,820 Like what's the point of cannabis data at the end of the day? 9 00:00:29,820 --> 00:00:32,040 So I'll get into that momentarily. 10 00:00:32,040 --> 00:00:38,920 But before you formally kick off, for those of you who are new, my name is Keegan. 11 00:00:38,920 --> 00:00:40,240 Started Cannlytics. 12 00:00:40,240 --> 00:00:45,880 So anyone can help out with statistics in the cannabis space. 13 00:00:45,880 --> 00:00:50,420 I think the cannabis space is a place for everyone. 14 00:00:50,420 --> 00:00:59,260 So grab your skills, grab any know-how you know, and come and join in on the fun. 15 00:00:59,260 --> 00:01:05,040 So the idea is it's not an exclusive club. 16 00:01:05,040 --> 00:01:10,560 Some industries are really hard to break into. 17 00:01:10,560 --> 00:01:20,000 And I think it would be real cool to see the cannabis industry be a place that's open to 18 00:01:20,000 --> 00:01:21,000 everybody. 19 00:01:21,000 --> 00:01:24,560 And so Cannlytics is leading by doing. 20 00:01:24,560 --> 00:01:28,500 And so Cannlytics is open to everyone. 21 00:01:28,500 --> 00:01:29,560 Any skill set. 22 00:01:29,560 --> 00:01:39,200 And so the idea is, I mean, myself, by and far, far from the greatest of whatever you 23 00:01:39,200 --> 00:01:48,880 may want to call one's self, whether it's a statistician, economist, data scientists. 24 00:01:48,880 --> 00:01:54,720 I may use those words, but people may disagree, you know, if I am any of those things. 25 00:01:54,720 --> 00:02:00,460 But the idea is I think that we can put all of our minds together with these awesome tools 26 00:02:00,460 --> 00:02:01,800 that we have here, right? 27 00:02:01,800 --> 00:02:04,080 The internet, computers. 28 00:02:04,080 --> 00:02:07,920 And I think we can make a positive impact to help a lot of people. 29 00:02:07,920 --> 00:02:15,040 So the idea is, you know, why should we just, you know, stand by and let all these golden 30 00:02:15,040 --> 00:02:17,240 opportunities get snatched up? 31 00:02:17,240 --> 00:02:22,280 So this is a place that I think should be open. 32 00:02:22,280 --> 00:02:27,120 And you know, anybody is welcome to come here, share your ideas, share your research. 33 00:02:27,120 --> 00:02:34,440 And the idea is by bouncing off our ideas together in an open manner where anyone can 34 00:02:34,440 --> 00:02:43,120 participate, values open to everybody, that we here can generate more value than you can 35 00:02:43,120 --> 00:02:47,920 with a treasure chest of six hundred million dollars. 36 00:02:47,920 --> 00:02:51,400 And I think we've already demonstrated that. 37 00:02:51,400 --> 00:02:54,560 And I would love to continue demonstrating that. 38 00:02:54,560 --> 00:02:59,320 And so I'm going to try to drive that point home here today. 39 00:02:59,320 --> 00:03:02,120 But that's what the group's about. 40 00:03:02,120 --> 00:03:09,400 And so without stealing the show, we'd like to spend today as a pretty open discussion 41 00:03:09,400 --> 00:03:11,480 so everyone has a chance to speak. 42 00:03:11,480 --> 00:03:13,640 So you don't have to. 43 00:03:13,640 --> 00:03:16,880 You're welcome to be a fly on the wall and listen. 44 00:03:16,880 --> 00:03:20,240 But if you want to speak up, speak up as much as you wish. 45 00:03:20,240 --> 00:03:26,680 So in no particular order, just starting in my top corner, Jacob, welcome to the group. 46 00:03:26,680 --> 00:03:32,120 Would you be interested in introducing yourself and what you hope to get out of the group 47 00:03:32,120 --> 00:03:43,680 or do in the canvas space? 48 00:03:43,680 --> 00:03:46,260 And no, no pressure, as I said. 49 00:03:46,260 --> 00:03:50,880 Or if you're taking a second to toggle your dials, I know it took me a second to then 50 00:03:50,880 --> 00:03:54,440 just chime in when you get those set. 51 00:03:54,440 --> 00:03:56,200 How about you, Candace? 52 00:03:56,200 --> 00:03:57,520 Anything on your mind? 53 00:03:57,520 --> 00:04:01,160 Anything that you're hoping to achieve this week? 54 00:04:01,160 --> 00:04:02,160 Thank you again. 55 00:04:02,160 --> 00:04:03,160 Let's see. 56 00:04:03,160 --> 00:04:10,800 I'm still working on the ERD, the markdown, and I did get the Panda profile reports for 57 00:04:10,800 --> 00:04:17,240 all the Washington State 2021 data set, and I'm just going to stuff it into a little PPT 58 00:04:17,240 --> 00:04:22,320 and we can start, you know, just kind of keep chunking away on that Washington State data 59 00:04:22,320 --> 00:04:25,160 set and who knows, you know, my call pretty cool. 60 00:04:25,160 --> 00:04:27,920 And then there's some other things. 61 00:04:27,920 --> 00:04:33,600 And then once we do some data analysis on it or get the data better, you know, we can 62 00:04:33,600 --> 00:04:40,000 start coming up with a bit of a plan and then I can, you know, put it up for everybody. 63 00:04:40,000 --> 00:04:42,280 And this is phenomenal. 64 00:04:42,280 --> 00:04:48,040 So a big complaint is that people trying to learn data science, they don't have good data 65 00:04:48,040 --> 00:04:52,440 sets to work on or they're all working on the same data sets, right? 66 00:04:52,440 --> 00:04:57,560 You all worked on the same Twitter posts and it's like, you know, how much can we scrape 67 00:04:57,560 --> 00:04:59,760 these Twitter posts? 68 00:04:59,760 --> 00:05:05,440 And I think that's one of the fun things about the Cannabis Data Science Group is where we 69 00:05:05,440 --> 00:05:08,000 have novel data sets. 70 00:05:08,000 --> 00:05:14,520 And we have to scrape them out of PDFs or APIs or what have you. 71 00:05:14,520 --> 00:05:15,520 But they're fun. 72 00:05:15,520 --> 00:05:16,520 It's real data. 73 00:05:16,520 --> 00:05:18,200 So we get real challenges. 74 00:05:18,200 --> 00:05:24,200 The Washington State data is one of the best examples of messy, messy data that a good 75 00:05:24,200 --> 00:05:30,000 data scientist needs, is needed to clean up and analyze. 76 00:05:30,000 --> 00:05:39,960 So your work on that is going to go light years because once you get that data set approachable, 77 00:05:39,960 --> 00:05:48,520 then people all across the world can use this, even if it's just as a test, right? 78 00:05:48,520 --> 00:05:54,360 So instead of testing your machine learning models on Twitter posts, you can test your 79 00:05:54,360 --> 00:06:00,840 machine learning models on Washington State traceability data, which is kind of cool in 80 00:06:00,840 --> 00:06:01,880 my opinion. 81 00:06:01,880 --> 00:06:04,540 So fantastic work, Candice. 82 00:06:04,540 --> 00:06:06,720 This will pay off here in the coming weeks. 83 00:06:06,720 --> 00:06:07,720 So thank you. 84 00:06:07,720 --> 00:06:12,360 Oh, and yeah, Ryan had put in a note and he said Canalytics. 85 00:06:12,360 --> 00:06:17,480 This one is C-A-N-N-L-Y-T-I-C-S, Ryan. 86 00:06:17,480 --> 00:06:20,080 But you can take it. 87 00:06:20,080 --> 00:06:21,080 Sorry, Keegan. 88 00:06:21,080 --> 00:06:22,080 Ah, so Canalytics. 89 00:06:22,080 --> 00:06:25,080 Yeah, Canalytics.ai.ai. 90 00:06:25,080 --> 00:06:28,840 So thank you. 91 00:06:28,840 --> 00:06:31,120 Oh, you're welcome. 92 00:06:31,120 --> 00:06:33,720 Thank you so much, Keegan, for what you're doing. 93 00:06:33,720 --> 00:06:37,720 And everybody, everybody that tips into Jerry, Joe, and everybody. 94 00:06:37,720 --> 00:06:41,320 It's everyone, I mean, awesome. 95 00:06:41,320 --> 00:06:42,320 Thank you. 96 00:06:42,320 --> 00:06:52,040 Candice, can I ask you a question about the Washington data set you're working with? 97 00:06:52,040 --> 00:06:54,480 Sure, John, if that's okay, Candice. 98 00:06:54,480 --> 00:06:56,720 Oh, yeah, sorry. 99 00:06:56,720 --> 00:06:59,440 Playing with, I guess, toggling the switches. 100 00:06:59,440 --> 00:07:00,440 Yeah. 101 00:07:00,440 --> 00:07:07,760 No, quick question is that our own organization has an interest in ingested products at this 102 00:07:07,760 --> 00:07:14,880 point and kind of typing the classes. 103 00:07:14,880 --> 00:07:22,600 We've done an initial pass with just survey data that we've collected from users in California 104 00:07:22,600 --> 00:07:28,160 that are patients of my clinical colleague, Jean Taleran. 105 00:07:28,160 --> 00:07:39,200 Does the Washington data set, does it have fields that speak to ingested data and ingested 106 00:07:39,200 --> 00:07:45,880 products and potentially we could look at that to try and figure out the type and frequency? 107 00:07:45,880 --> 00:07:53,280 Well, they're not doing like per customer dosing per se. 108 00:07:53,280 --> 00:07:59,800 No, no, but do they talk about gummies versus chocolates versus mints versus beverages? 109 00:07:59,800 --> 00:08:07,680 Yes, and Keegan has used that data set to in past YouTube videos showing that too, I 110 00:08:07,680 --> 00:08:13,640 think, the way with Washington State, right, we've done Washington State, Oklahoma, all 111 00:08:13,640 --> 00:08:15,440 kinds of data sets. 112 00:08:15,440 --> 00:08:18,000 But yes, the answer is yes. 113 00:08:18,000 --> 00:08:20,440 There's all different types of flour. 114 00:08:20,440 --> 00:08:21,440 There's edibles. 115 00:08:21,440 --> 00:08:23,800 There's, I don't think they actually break it down. 116 00:08:23,800 --> 00:08:25,160 Keegan, correct me if I'm wrong. 117 00:08:25,160 --> 00:08:32,040 I don't think they break it down into like whether they're chocolates, gummies, but they 118 00:08:32,040 --> 00:08:37,640 do kind of, yeah, they do indicate that. 119 00:08:37,640 --> 00:08:44,680 Well, many thoughts because you two both raised many good ideas here and points. 120 00:08:44,680 --> 00:08:46,240 So where to begin? 121 00:08:46,240 --> 00:08:51,600 I would begin by saying once you find one gold nugget in a data set, then there could 122 00:08:51,600 --> 00:08:53,100 easily be more. 123 00:08:53,100 --> 00:08:58,800 So as you raised out, I raised the point, there's probably more gold in the Washington 124 00:08:58,800 --> 00:09:00,640 State data. 125 00:09:00,640 --> 00:09:08,080 Sure enough, John, I think we can approach your question because we can look at solid 126 00:09:08,080 --> 00:09:10,140 and liquid edibles. 127 00:09:10,140 --> 00:09:16,500 If we wanted to get fancy, we could use natural language processing and try to figure out 128 00:09:16,500 --> 00:09:20,520 which ones are chocolates, which ones are gummies, so on and so forth. 129 00:09:20,520 --> 00:09:23,080 But that would be a cherry on top. 130 00:09:23,080 --> 00:09:26,360 And then as you said, so basically, John's after the types. 131 00:09:26,360 --> 00:09:30,200 So this would be THC to CBD ratio. 132 00:09:30,200 --> 00:09:32,760 And sure enough, we've got that data. 133 00:09:32,760 --> 00:09:38,500 So with the 2022 data, we still have to make that link. 134 00:09:38,500 --> 00:09:42,200 So we're still missing some IDs that we still need to request. 135 00:09:42,200 --> 00:09:51,280 But up to the end of 2021, we can match lab results to edibles that were sold. 136 00:09:51,280 --> 00:10:00,640 So we can see the distribution of THC to CBD in edibles that were sold in Washington State 137 00:10:00,640 --> 00:10:03,440 of the full population. 138 00:10:03,440 --> 00:10:06,220 Excellent. 139 00:10:06,220 --> 00:10:10,560 So easier said than done, wrangling this data. 140 00:10:10,560 --> 00:10:15,200 So Candice, now you actually have a demanded statistic. 141 00:10:15,200 --> 00:10:24,080 So we now want to know the THC to CBD ratio, which is a statistic for every single edible 142 00:10:24,080 --> 00:10:25,840 that was sold. 143 00:10:25,840 --> 00:10:29,280 And then we can subsequently create more statistics. 144 00:10:29,280 --> 00:10:41,920 So you can get fancy from there, but maybe there's a pattern in rural versus urban or 145 00:10:41,920 --> 00:10:42,920 who knows. 146 00:10:42,920 --> 00:10:48,440 So that's where you can start getting fun thinking of research questions. 147 00:10:48,440 --> 00:10:51,440 But first things first, we'll get the first statistic. 148 00:10:51,440 --> 00:10:53,880 So Candice, you want to start on that? 149 00:10:53,880 --> 00:10:54,880 You're welcome to. 150 00:10:54,880 --> 00:10:58,440 What do you mean on the first statistic that we have? 151 00:10:58,440 --> 00:11:03,600 Are you on the THC ratio for all Washington State data 2021 edibles? 152 00:11:03,600 --> 00:11:06,400 Is that what you said? 153 00:11:06,400 --> 00:11:08,920 Essentially, yeah, we'll have to jump through a couple hoops. 154 00:11:08,920 --> 00:11:10,720 We'll have to get all the edibles. 155 00:11:10,720 --> 00:11:11,720 Yeah. 156 00:11:11,720 --> 00:11:14,800 Oh yeah, yeah, we'll get it done. 157 00:11:14,800 --> 00:11:23,280 And then essentially just take the maybe total THC divided by total CBD. 158 00:11:23,280 --> 00:11:26,400 Sounds good. 159 00:11:26,400 --> 00:11:37,120 If I might just make a point here, if the data contains non-THC, non-CBD cannabinoid 160 00:11:37,120 --> 00:11:44,720 content fields, that would be exactly the goal that we're trying to approach. 161 00:11:44,720 --> 00:11:51,120 So maybe when this moves to a point where you're willing to present this, maybe that 162 00:11:51,120 --> 00:11:57,400 would be a good opportunity to see what can be mined from it. 163 00:11:57,400 --> 00:12:01,560 So what you're talking about are like other ones like CBN, CBG. 164 00:12:01,560 --> 00:12:05,040 Yeah, CBG, CBN, THCV. 165 00:12:05,040 --> 00:12:09,520 If that in fact are fields in the Washington State data. 166 00:12:09,520 --> 00:12:15,120 Right, or like the fields that are in skunk effects. 167 00:12:15,120 --> 00:12:17,640 I'm sorry, say that last one again, Candice. 168 00:12:17,640 --> 00:12:19,440 Like the fields in skunk effects. 169 00:12:19,440 --> 00:12:21,240 Can you hear me okay? 170 00:12:21,240 --> 00:12:22,240 Yeah, yeah, yeah. 171 00:12:22,240 --> 00:12:23,240 Am I too loud? 172 00:12:23,240 --> 00:12:24,240 No, no, you're perfect. 173 00:12:24,240 --> 00:12:30,960 Okay, I did test. 174 00:12:30,960 --> 00:12:35,640 Just not to rain on your parade, but I think you may have the best luck with the THC and 175 00:12:35,640 --> 00:12:45,800 CBD because I mean it would be fun to see if there are any CBG, etc. edibles, but just 176 00:12:45,800 --> 00:12:51,800 from my experience seeing a bunch of edible lab results, I just don't... 177 00:12:51,800 --> 00:13:02,640 I mean the predominant ones you see of course are CBD, CBDA, THC, THCA. 178 00:13:02,640 --> 00:13:04,840 Wait a minute, repeat that real quick. 179 00:13:04,840 --> 00:13:09,560 So okay, CBDA, CBN, CBG. 180 00:13:09,560 --> 00:13:16,000 So basically you should expect to see CBD, CBDA. 181 00:13:16,000 --> 00:13:20,360 CBDA, okay, got that. 182 00:13:20,360 --> 00:13:24,000 THC, and then THCA. 183 00:13:24,000 --> 00:13:26,880 THCA, okay. 184 00:13:26,880 --> 00:13:29,720 So you'd expect to see... 185 00:13:29,720 --> 00:13:31,220 And then THCB. 186 00:13:31,220 --> 00:13:36,600 You would expect to see mostly THC and CBD. 187 00:13:36,600 --> 00:13:41,280 It would be curious to see if there's any acidic left. 188 00:13:41,280 --> 00:13:52,080 So CBDA and THCA, I would expect to be zero because the idea is when you process it into 189 00:13:52,080 --> 00:14:00,280 an edible, you typically have it go through decarboxylation. 190 00:14:00,280 --> 00:14:04,440 So you turn the acidic to their... 191 00:14:04,440 --> 00:14:10,040 John, you probably know the correct word for this. 192 00:14:10,040 --> 00:14:12,360 Carboxylated or neutral form. 193 00:14:12,360 --> 00:14:14,980 So they're psychoactive at that point. 194 00:14:14,980 --> 00:14:16,720 So you would expect... 195 00:14:16,720 --> 00:14:21,160 But I mean they still may have trace amounts of CBDA and THCA. 196 00:14:21,160 --> 00:14:29,560 But long story short is these are typically ingredients that would be added as isolates 197 00:14:29,560 --> 00:14:35,680 or distillates into a soda or into an edible. 198 00:14:35,680 --> 00:14:49,200 So it would be unlikely for somebody to be adding in CBG, THCV, CBN, et cetera, but they 199 00:14:49,200 --> 00:14:50,200 may. 200 00:14:50,200 --> 00:14:51,200 So for example... 201 00:14:51,200 --> 00:14:53,240 And Keegan, they are. 202 00:14:53,240 --> 00:14:58,600 They're rarer products and that's why it becomes of interest because we're trying to tease 203 00:14:58,600 --> 00:15:00,600 out the story on those. 204 00:15:00,600 --> 00:15:02,480 No, they definitely exist. 205 00:15:02,480 --> 00:15:06,920 We have them and others have them in their product databases that we've been looking 206 00:15:06,920 --> 00:15:07,920 at. 207 00:15:07,920 --> 00:15:12,600 But as you point out, of course, it's just rarer because the market hasn't moved in that 208 00:15:12,600 --> 00:15:14,680 direction yet so much. 209 00:15:14,680 --> 00:15:20,480 But it is a very fruitful, interesting area and there are big players looking at this. 210 00:15:20,480 --> 00:15:23,920 So let's just keep an eye on... 211 00:15:23,920 --> 00:15:26,600 They're called rare cannabinoids for a reason. 212 00:15:26,600 --> 00:15:28,600 And you know what else is rare? 213 00:15:28,600 --> 00:15:29,600 Gold. 214 00:15:29,600 --> 00:15:31,920 And so that's basically what we're doing. 215 00:15:31,920 --> 00:15:34,080 We're panning for gold here. 216 00:15:34,080 --> 00:15:36,440 So we've got the tools. 217 00:15:36,440 --> 00:15:41,160 So basically we can look through this data set and see if are there any edibles that 218 00:15:41,160 --> 00:15:44,240 were sold with CBN. 219 00:15:44,240 --> 00:15:52,760 So I think I've even seen some at the store and basically I guess it's not really a claim, 220 00:15:52,760 --> 00:16:01,920 but the link that's trying to be advertised is maybe CBN maybe correlated with a sleepy 221 00:16:01,920 --> 00:16:03,920 effect. 222 00:16:03,920 --> 00:16:10,480 So I don't know how people are marketing the compounds, but this is something that we can 223 00:16:10,480 --> 00:16:11,480 investigate. 224 00:16:11,480 --> 00:16:13,880 So we've got some cool tools in our belt. 225 00:16:13,880 --> 00:16:19,580 So what I would do is look at natural language processing plus these rare compounds. 226 00:16:19,580 --> 00:16:27,040 So try to find any edibles that have CBN, etc. in them. 227 00:16:27,040 --> 00:16:33,640 Use natural language processing on the sample name and try to figure out are they capsules? 228 00:16:33,640 --> 00:16:38,120 Are they chocolates? 229 00:16:38,120 --> 00:16:39,120 So on and so forth. 230 00:16:39,120 --> 00:16:47,280 So that's where you can get fun with it to see if there's any insights that you could 231 00:16:47,280 --> 00:16:48,280 pull out. 232 00:16:48,280 --> 00:16:52,600 And then finally I think the business question is what's selling? 233 00:16:52,600 --> 00:17:00,480 So if you can find some CBN edibles that are selling off the shelf in Seattle, then that 234 00:17:00,480 --> 00:17:03,120 would be a valuable find. 235 00:17:03,120 --> 00:17:08,400 And the data is just sitting there and I don't know if anybody's looked at that, at least 236 00:17:08,400 --> 00:17:10,200 not publicly. 237 00:17:10,200 --> 00:17:15,040 And so now we've got a bunch of awesome data scientists here. 238 00:17:15,040 --> 00:17:17,600 The question's been posed. 239 00:17:17,600 --> 00:17:20,680 We've got all the open source tools in our tool belt. 240 00:17:20,680 --> 00:17:24,400 So I think we can answer this question for you. 241 00:17:24,400 --> 00:17:36,160 So a coming question or topic I think for next week is so how do we get compensated 242 00:17:36,160 --> 00:17:37,480 for our work? 243 00:17:37,480 --> 00:17:49,720 And a nifty tool that I'll be sharing with you is essentially... 244 00:17:49,720 --> 00:17:51,760 Okay I'll go ahead and let the cat out of the bag. 245 00:17:51,760 --> 00:18:02,840 So basically what we're looking at here is basically the word has gotten a little overused, 246 00:18:02,840 --> 00:18:04,240 a little hyped. 247 00:18:04,240 --> 00:18:14,880 But the idea is we can basically create non-fungible tokens, NFTs, for data as well as algorithms. 248 00:18:14,880 --> 00:18:25,600 So the idea is we could all write little bitty algorithms that help collect data. 249 00:18:25,600 --> 00:18:31,880 You would have ownership of the algorithms you wrote and you could potentially buy or 250 00:18:31,880 --> 00:18:32,880 sell them. 251 00:18:32,880 --> 00:18:37,400 And then people could then pay you to use your algorithms. 252 00:18:37,400 --> 00:18:43,120 And so the idea is Candice may write a nifty algorithm. 253 00:18:43,120 --> 00:18:47,240 I may use another algorithm. 254 00:18:47,240 --> 00:18:53,640 Maybe that incorporates Candice's and maybe I pay her a fee to use hers. 255 00:18:53,640 --> 00:18:55,200 Or maybe I write my own. 256 00:18:55,200 --> 00:19:00,160 And then maybe John wants to come along and then use those two algorithms to get some 257 00:19:00,160 --> 00:19:01,160 data. 258 00:19:01,160 --> 00:19:07,680 And so the idea is John could pay a nominal fee to Candice to use her algorithm or to 259 00:19:07,680 --> 00:19:08,680 me. 260 00:19:08,680 --> 00:19:12,560 Hopefully the fee's low enough that it's well worth John's while. 261 00:19:12,560 --> 00:19:15,840 So pennies on the dollar is the idea. 262 00:19:15,840 --> 00:19:20,480 So then Candice gets compensated for writing the algorithm. 263 00:19:20,480 --> 00:19:22,800 John gets the data he needs. 264 00:19:22,800 --> 00:19:25,600 It's all perfectly traceable. 265 00:19:25,600 --> 00:19:33,720 So there's no question about where did this data come from, how is the data cleaned. 266 00:19:33,720 --> 00:19:40,320 It's all essentially on the blockchain. 267 00:19:40,320 --> 00:19:42,720 We can decide which blockchain. 268 00:19:42,720 --> 00:19:50,960 So I'll talk specifics about this more next week because this is something I've... 269 00:19:50,960 --> 00:19:53,040 I'm standing on the shoulders of giants. 270 00:19:53,040 --> 00:19:57,720 So this is a project that somebody's already done here. 271 00:19:57,720 --> 00:20:00,880 So long story short, this is a project that somebody's already done. 272 00:20:00,880 --> 00:20:04,680 We're just wrapping it up into Canlittix. 273 00:20:04,680 --> 00:20:09,600 So if you're particularly interested in this, let me know because I'll share more about 274 00:20:09,600 --> 00:20:11,240 the project details with you. 275 00:20:11,240 --> 00:20:15,120 I'm trying to keep it a little hush hush. 276 00:20:15,120 --> 00:20:20,480 But at the same time, I think because we're this open group, I don't think closed groups 277 00:20:20,480 --> 00:20:21,640 can compete with us. 278 00:20:21,640 --> 00:20:25,080 So that's why I'm fairly open about our ideas. 279 00:20:25,080 --> 00:20:31,720 So for example, this SkunkFX model that we just built, we built that in a matter of two 280 00:20:31,720 --> 00:20:33,280 weeks or so. 281 00:20:33,280 --> 00:20:35,760 And yes, it's got its flaws. 282 00:20:35,760 --> 00:20:38,040 Yes, it can be improved. 283 00:20:38,040 --> 00:20:42,360 But this is a model that would have taken a corporation. 284 00:20:42,360 --> 00:20:43,360 What? 285 00:20:43,360 --> 00:20:47,500 I mean, they would have waited to file the trademark. 286 00:20:47,500 --> 00:20:50,320 They would have waited to file a patent. 287 00:20:50,320 --> 00:20:56,880 This would have been six months to two years out the door easily and millions of dollars 288 00:20:56,880 --> 00:20:59,240 spent on what? 289 00:20:59,240 --> 00:21:07,480 And so here we just effortlessly made a prediction model in a matter of weeks that companies 290 00:21:07,480 --> 00:21:09,880 just can't touch. 291 00:21:09,880 --> 00:21:16,520 So that's why it's been sort of slow demonstrating the power of Canlittix. 292 00:21:16,520 --> 00:21:22,360 And I mean, we have all the greatest data scientists in the world on this. 293 00:21:22,360 --> 00:21:26,400 And the idea is now we just need to get them compensated. 294 00:21:26,400 --> 00:21:33,840 And that's exactly what the data NFTs, the algorithm NFTs bring to the picture. 295 00:21:33,840 --> 00:21:40,980 It finally adds a mechanism for all of you great contributors to finally get paid. 296 00:21:40,980 --> 00:21:44,320 So now you can get paid for your algorithms. 297 00:21:44,320 --> 00:21:46,560 You can get paid for finding data. 298 00:21:46,560 --> 00:21:49,440 You can get paid for sharing data. 299 00:21:49,440 --> 00:21:55,920 And I think this has been a critical missing piece of the puzzle. 300 00:21:55,920 --> 00:21:59,080 So this will be getting built in the coming weeks. 301 00:21:59,080 --> 00:22:03,160 So it's sort of a call to action right now. 302 00:22:03,160 --> 00:22:11,860 So if any of you want to help contribute, then at the moment can add you on to authorship. 303 00:22:11,860 --> 00:22:13,880 You can share in the copyright. 304 00:22:13,880 --> 00:22:19,620 So at the moment, we can share copyrights and get this built. 305 00:22:19,620 --> 00:22:27,760 And then at that point, you can actually get paid for writing algorithms or cleaning data. 306 00:22:27,760 --> 00:22:31,200 And I think this is going to be revolutionary. 307 00:22:31,200 --> 00:22:35,120 So as I said, this is a project that somebody is just doing in general. 308 00:22:35,120 --> 00:22:42,180 It's just general data NFTs, general algorithm NFTs. 309 00:22:42,180 --> 00:22:47,460 And I think we can get it to catch on here in the cannabis space. 310 00:22:47,460 --> 00:22:52,120 And I foresee things like this catching on in most fields. 311 00:22:52,120 --> 00:22:57,160 And then that way, data scientists like yourself could actually get compensated. 312 00:22:57,160 --> 00:22:59,400 So I think it's a cool project. 313 00:22:59,400 --> 00:23:02,560 We'll just be an example of it. 314 00:23:02,560 --> 00:23:06,560 Keegan, can I just follow on on that? 315 00:23:06,560 --> 00:23:13,400 Because as this gets discussed within this group, I would just point out that there's 316 00:23:13,400 --> 00:23:21,040 an organization called the Ethical Data Alliance, which is absolutely doing this, has some grants 317 00:23:21,040 --> 00:23:28,000 from some IT organizations to help move this along. 318 00:23:28,000 --> 00:23:33,240 It's in the cannabis space because that's their focus. 319 00:23:33,240 --> 00:23:37,720 And they've been talking about or they are creating something called the Ethical Eden, 320 00:23:37,720 --> 00:23:41,940 whatever that is, the network that this would be on. 321 00:23:41,940 --> 00:23:51,320 So I would just say it's probably worth looking at or talking to folks in EDA just to see 322 00:23:51,320 --> 00:23:57,160 how it overlaps or complements or whatever. 323 00:23:57,160 --> 00:23:58,280 This is going on. 324 00:23:58,280 --> 00:24:01,760 And so it's worth being aware of it. 325 00:24:01,760 --> 00:24:03,240 Exactly. 326 00:24:03,240 --> 00:24:09,720 And I kind of want to tie this into something where I often say here, oh, you know, we're 327 00:24:09,720 --> 00:24:14,480 advancing cannabis science and we're advancing statistics and this and that. 328 00:24:14,480 --> 00:24:20,120 Does that mean that on the greatest mine out there, advancing the greatest theories, doing 329 00:24:20,120 --> 00:24:21,640 the greatest work? 330 00:24:21,640 --> 00:24:25,160 No, no, no, no, no, no, far from it. 331 00:24:25,160 --> 00:24:31,240 That was sort of the point at the very beginning is we may be some of the least skilled. 332 00:24:31,240 --> 00:24:35,160 And we're just doing the best we can. 333 00:24:35,160 --> 00:24:42,680 But the idea is if we get a bunch of small fish together, we can form a giant school 334 00:24:42,680 --> 00:24:46,040 of fish. 335 00:24:46,040 --> 00:24:49,920 And the whole idea is stand on the shoulders of giants. 336 00:24:49,920 --> 00:24:55,480 So the idea is these are awesome tools that some of the brightest minds in the world are 337 00:24:55,480 --> 00:24:56,480 making. 338 00:24:56,480 --> 00:24:59,440 We don't necessarily need to make them. 339 00:24:59,440 --> 00:25:04,720 If we use these tools, that helps advance the field. 340 00:25:04,720 --> 00:25:11,600 So in my opinion, so if we use statistics, I think that helps advance statistics. 341 00:25:11,600 --> 00:25:18,080 If we apply statistical models to the cannabis industry in a new way that nobody else has 342 00:25:18,080 --> 00:25:21,440 done before, I think that helps advance the fields. 343 00:25:21,440 --> 00:25:23,600 Are our statistics perfect? 344 00:25:23,600 --> 00:25:24,600 No. 345 00:25:24,600 --> 00:25:26,680 Does the analysis need to be done again? 346 00:25:26,680 --> 00:25:27,680 Yes. 347 00:25:27,680 --> 00:25:29,880 Does the analysis need to be done more rigorously? 348 00:25:29,880 --> 00:25:30,880 Yes. 349 00:25:30,880 --> 00:25:32,880 Do we need to do more homework? 350 00:25:32,880 --> 00:25:33,880 Yes. 351 00:25:33,880 --> 00:25:35,960 Are we asking novel questions? 352 00:25:35,960 --> 00:25:37,720 I hope so. 353 00:25:37,720 --> 00:25:40,680 Are we applying methods in novel ways? 354 00:25:40,680 --> 00:25:41,920 I hope so. 355 00:25:41,920 --> 00:25:48,280 So I think that's sort of the idea behind the contribution we make is somebody needs 356 00:25:48,280 --> 00:25:51,800 to come back and do it much, much better. 357 00:25:51,800 --> 00:25:54,840 But it's sort of like the law of large numbers, right? 358 00:25:54,840 --> 00:26:01,880 You need to do a scientific test, right, thousands and thousands of times. 359 00:26:01,880 --> 00:26:05,800 But the first one provides a lot of value. 360 00:26:05,800 --> 00:26:08,400 And it could be wildly, wildly off. 361 00:26:08,400 --> 00:26:10,440 But it gets people thinking. 362 00:26:10,440 --> 00:26:13,480 And then somebody can come along and test it again. 363 00:26:13,480 --> 00:26:16,040 And then test it again, test it again, test it again. 364 00:26:16,040 --> 00:26:20,400 And so what is the actual result? 365 00:26:20,400 --> 00:26:23,280 Easily, could change over time. 366 00:26:23,280 --> 00:26:27,000 So as I said, take all of our analysis as a grain of salt. 367 00:26:27,000 --> 00:26:35,040 But I think the value we add is we may introduce an idea into the space, at which point better 368 00:26:35,040 --> 00:26:39,000 and greater minds can iterate upon this. 369 00:26:39,000 --> 00:26:46,920 So that's what I mean when I say I think we're advancing cannabis science is I think we're 370 00:26:46,920 --> 00:26:49,800 just treading on ground that no one's gone before. 371 00:26:49,800 --> 00:26:52,400 And I think there's value to that. 372 00:26:52,400 --> 00:26:56,600 But we do need to do so in an ethical manner, right? 373 00:26:56,600 --> 00:27:02,000 When you're treading on new ground, you can't just be leaving a bunch of garbage all over 374 00:27:02,000 --> 00:27:09,360 the place or making a mess for people who may wish to follow in your footsteps. 375 00:27:09,360 --> 00:27:15,400 So we do need to tread carefully. 376 00:27:15,400 --> 00:27:17,560 So that's getting a little abstract. 377 00:27:17,560 --> 00:27:25,960 So if you're interested, I've got a quick 15 minute presentation to share with you. 378 00:27:25,960 --> 00:27:29,840 And then John, do you have any work to share or? 379 00:27:29,840 --> 00:27:33,720 I have a couple points I wouldn't mind bringing up to the group at this point. 380 00:27:33,720 --> 00:27:38,760 So sure, but if there's time, would you mind? 381 00:27:38,760 --> 00:27:46,120 Well, would you mind if I ran through just a please? 382 00:27:46,120 --> 00:27:53,560 Well, actually, actually, if you're OK, I'll just run through just a teaser of the work 383 00:27:53,560 --> 00:27:55,520 I'll be working on. 384 00:27:55,520 --> 00:28:01,880 And if any of you want to lend a hand and then, John, I'll give it the floor to you 385 00:28:01,880 --> 00:28:04,800 for the last 15 minutes or so for you to. 386 00:28:04,800 --> 00:28:06,520 I don't even need that much. 387 00:28:06,520 --> 00:28:14,520 OK, well, without further ado, I'll just share with you real, real quick what I had been 388 00:28:14,520 --> 00:28:16,320 thinking about. 389 00:28:16,320 --> 00:28:22,240 So this you can think of this what it was, but this is sort of the point about canlytics 390 00:28:22,240 --> 00:28:24,800 is what do I value? 391 00:28:24,800 --> 00:28:31,600 So this is just a fictitious quote that I heard one time that I really liked a lot. 392 00:28:31,600 --> 00:28:37,800 And so to me, I personally think code is maybe one of the most valuable things out there. 393 00:28:37,800 --> 00:28:40,760 So that's what I value. 394 00:28:40,760 --> 00:28:43,720 And that's what canlytics values. 395 00:28:43,720 --> 00:28:49,560 So if any of you want to share in the copyright of the code, I think that's a valuable thing. 396 00:28:49,560 --> 00:28:53,000 And so I'm more than happy to share that with you. 397 00:28:53,000 --> 00:28:59,460 So please, please email me if you have any contribution and a contribution can mean anything. 398 00:28:59,460 --> 00:29:05,240 So really, you know, if you help help share ideas or share data sets with me, I would 399 00:29:05,240 --> 00:29:07,280 love to share authorship with you. 400 00:29:07,280 --> 00:29:10,040 And I haven't been very good about that in the past. 401 00:29:10,040 --> 00:29:14,640 So please be in touch with me and I'll be happy to share authorship with people. 402 00:29:14,640 --> 00:29:15,640 Abundance. 403 00:29:15,640 --> 00:29:23,700 Next, going to be rigorous here about following the scientific process. 404 00:29:23,700 --> 00:29:29,300 So it's just going to outline real quick the steps of a quick research question. 405 00:29:29,300 --> 00:29:36,120 So basically, the research question that that I posed, but I kind of want to formally look 406 00:29:36,120 --> 00:29:43,800 through it in a scientific manner is I've got this hypothesis that a cannabis consumer's 407 00:29:43,800 --> 00:29:50,560 personality affects the type of cannabis that they choose to purchase. 408 00:29:50,560 --> 00:30:00,680 So just to kind of just, you know, stab in the dark, I was thinking that, oh, you know, 409 00:30:00,680 --> 00:30:08,120 maybe just, you know, a hypothesis is maybe a little more than a guess, right? 410 00:30:08,120 --> 00:30:17,400 It's a guess or, you know, my belief from just what I've observed in the world. 411 00:30:17,400 --> 00:30:21,680 And basically, that right, there's introverts and extroverts. 412 00:30:21,680 --> 00:30:23,960 It's a bit of a scale. 413 00:30:23,960 --> 00:30:26,560 It's not a zero or one thing. 414 00:30:26,560 --> 00:30:31,680 So we're expecting a scale and there's other personality traits, but that's maybe one of 415 00:30:31,680 --> 00:30:33,160 them. 416 00:30:33,160 --> 00:30:37,300 And maybe they like cannabis differently. 417 00:30:37,300 --> 00:30:45,640 So I've got this hunch that maybe extroverts may like more of the indica type strains. 418 00:30:45,640 --> 00:30:52,560 And then maybe the introverts may like more of the sativa like strains. 419 00:30:52,560 --> 00:30:54,320 Is there anything to that? 420 00:30:54,320 --> 00:30:55,540 Who knows? 421 00:30:55,540 --> 00:30:59,640 We can kind of test it statistically. 422 00:30:59,640 --> 00:31:01,800 So how do we go about doing that? 423 00:31:01,800 --> 00:31:09,960 Well, first, we need to look at the literature to see who's talked about this before and 424 00:31:09,960 --> 00:31:16,440 what concepts in the scientific literature can we use. 425 00:31:16,440 --> 00:31:26,280 And the big one I wanted to use is one, revealed preference, in that an economist would say 426 00:31:26,280 --> 00:31:34,280 you can't just ask somebody about themselves, about their preferences. 427 00:31:34,280 --> 00:31:37,640 You have to determine that from looking at their actions. 428 00:31:37,640 --> 00:31:44,720 So the idea here is we can't just ask somebody what their personality is. 429 00:31:44,720 --> 00:31:51,800 We have to determine what their personality is from their actions. 430 00:31:51,800 --> 00:31:53,440 Same with purchasing patterns. 431 00:31:53,440 --> 00:31:56,720 We can't just ask people what they want to purchase. 432 00:31:56,720 --> 00:32:00,160 We actually have to observe what they purchase. 433 00:32:00,160 --> 00:32:07,040 Next, what are we even talking about when we're talking about personality? 434 00:32:07,040 --> 00:32:13,640 Well, this is cool because now is where we kind of get the sample from different fields. 435 00:32:13,640 --> 00:32:22,320 So we get to sample from psychology and say, cool, the psychologists have studied personality 436 00:32:22,320 --> 00:32:31,320 and they've widely agreed upon five personality traits that kind of are on a scale. 437 00:32:31,320 --> 00:32:34,360 So I'm treating it as zero to one. 438 00:32:34,360 --> 00:32:46,160 So five traits on a scale, widely criticized and also widely referenced by psychologists. 439 00:32:46,160 --> 00:32:47,160 Cool. 440 00:32:47,160 --> 00:32:54,840 So we've got some shoulders of giants to stand upon. 441 00:32:54,840 --> 00:32:59,600 Next, we need our actual methodology. 442 00:32:59,600 --> 00:33:06,440 So what's our actual statistical framework that we'll be using? 443 00:33:06,440 --> 00:33:15,280 So I picked the simplest statistical model in the book, the ordinary least squares regression. 444 00:33:15,280 --> 00:33:22,700 I suppose difference of means or something like that may technically be a simpler method 445 00:33:22,700 --> 00:33:26,800 of analysis, but this is pretty simple. 446 00:33:26,800 --> 00:33:33,160 The idea here is how are we even measuring indica versus sativa? 447 00:33:33,160 --> 00:33:42,520 Well, thanks to awesome work being done by John, it's appearing that the ratio between 448 00:33:42,520 --> 00:33:51,880 beta-pinene and d-limonene may better characterize the sativa indica dichotomy. 449 00:33:51,880 --> 00:34:02,120 So instead of using a dichotomy, we'll use a continuous variable R, which is going to 450 00:34:02,120 --> 00:34:07,840 be the ratio between beta-pinene and d-limonene. 451 00:34:07,840 --> 00:34:10,680 We'll take the average by user. 452 00:34:10,680 --> 00:34:17,920 So maybe a user may pick a bunch of different strains, but maybe on average they'll pick 453 00:34:17,920 --> 00:34:21,600 out hopefully strains that they'd like. 454 00:34:21,600 --> 00:34:27,960 Or they may accidentally pick out a strain that's high in d-limonene and not like it. 455 00:34:27,960 --> 00:34:33,620 Or they may accidentally pick out a strain that's high in beta-pinene, love it, not like 456 00:34:33,620 --> 00:34:35,000 it, who knows? 457 00:34:35,000 --> 00:34:42,280 The idea is after they try many, many, many samples, they'll on average hopefully be picking 458 00:34:42,280 --> 00:34:45,780 out the ones that they like. 459 00:34:45,780 --> 00:34:48,040 Will this hold up to the data? 460 00:34:48,040 --> 00:34:49,840 We'll see. 461 00:34:49,840 --> 00:35:00,680 And then the idea is we can now have the five personality traits, openness, conscientiousness, 462 00:35:00,680 --> 00:35:06,280 extraversion, agreeableness, and neuroticism, which is basically sensitivity. 463 00:35:06,280 --> 00:35:14,160 I think they called it emotional sensitivity or something like that. 464 00:35:14,160 --> 00:35:16,360 Closed to new ideas is neuroticism. 465 00:35:16,360 --> 00:35:18,920 Say that one more time. 466 00:35:18,920 --> 00:35:23,000 Closed to new ideas. 467 00:35:23,000 --> 00:35:26,600 The emotional one, the neuroticism? 468 00:35:26,600 --> 00:35:34,360 I think the comparison is openness or closeness to new ideas. 469 00:35:34,360 --> 00:35:36,520 And I think that's what falls into neurotic. 470 00:35:36,520 --> 00:35:40,880 It's considered closure, a lack of openness. 471 00:35:40,880 --> 00:35:44,080 So I've got to do more research myself. 472 00:35:44,080 --> 00:35:48,640 I think closeness would be the converse to openness. 473 00:35:48,640 --> 00:35:52,680 I think this is basically emotional stability. 474 00:35:52,680 --> 00:36:00,560 So are you highly emotional or are you stoic? 475 00:36:00,560 --> 00:36:07,720 Do you let your emotions show or not is the neuroticism one, is the way I understand that 476 00:36:07,720 --> 00:36:10,240 one. 477 00:36:10,240 --> 00:36:18,360 And then agreeableness, agreeable, disagreeable, extraversion, introversion. 478 00:36:18,360 --> 00:36:26,480 I don't really know what the, I think the conscientious versus intuitive. 479 00:36:26,480 --> 00:36:31,680 So is this, like I said, I don't know too much about this. 480 00:36:31,680 --> 00:36:35,560 This one may be the logic versus feeling one. 481 00:36:35,560 --> 00:36:42,040 But anywho, please read up on these because this is a deep, deep subject, right? 482 00:36:42,040 --> 00:36:49,480 I mean, this is the field of psychology, so there's a lot of meat here. 483 00:36:49,480 --> 00:36:55,720 But the idea is if we scale these zero to one, right? 484 00:36:55,720 --> 00:36:56,720 Probably no one's... 485 00:36:56,720 --> 00:36:57,720 I have a question about that. 486 00:36:57,720 --> 00:37:03,440 If you go in and take all of those personality types and scale it zero to one, you should 487 00:37:03,440 --> 00:37:06,960 have two words there and not just one term. 488 00:37:06,960 --> 00:37:16,520 It should be openness, closeness, conscientiousness, whatever the opposite of that is. 489 00:37:16,520 --> 00:37:17,840 You're 100% right. 490 00:37:17,840 --> 00:37:24,120 And so I just grabbed these from Wikipedia, but please double check me on this because 491 00:37:24,120 --> 00:37:33,160 the idea is these words are supposed to convey a scale. 492 00:37:33,160 --> 00:37:39,960 It's basically like zero to one extraversion, zero to one agreeableness. 493 00:37:39,960 --> 00:37:46,440 So would zero be introversion or extraversion and one, which would one be? 494 00:37:46,440 --> 00:37:47,440 Good point. 495 00:37:47,440 --> 00:37:55,200 I'm basically treating one as extroverted and zero as introverted. 496 00:37:55,200 --> 00:38:05,440 Can I just make a comment here because this is precisely what we're facing as we talk 497 00:38:05,440 --> 00:38:10,320 about these kinds of scales in our dosing project work. 498 00:38:10,320 --> 00:38:17,120 We have a really great collaborator who's a clinical psychologist in the form of Jeffrey 499 00:38:17,120 --> 00:38:20,040 Terrence out of Eugene, Oregon. 500 00:38:20,040 --> 00:38:22,800 And so we do spend time talking about this. 501 00:38:22,800 --> 00:38:27,760 And point of fact, in some of our most recent discussions, he says, no, don't go binary 502 00:38:27,760 --> 00:38:29,600 on this. 503 00:38:29,600 --> 00:38:33,560 It's better to do it as a bit of an ordinal scale. 504 00:38:33,560 --> 00:38:37,320 It's got a term, I think it's called Langford or something, scale. 505 00:38:37,320 --> 00:38:44,640 And he's convinced us that as we do survey work, we need to not be doing it binary, but 506 00:38:44,640 --> 00:38:51,560 in fact asking folks for if they're responding in the moment or whatever, but it needs to 507 00:38:51,560 --> 00:38:56,760 be kind of a scale, Jerry, kind of along the way you're saying. 508 00:38:56,760 --> 00:38:57,760 I went into this. 509 00:38:57,760 --> 00:38:59,280 I'm so impressed with the binary. 510 00:38:59,280 --> 00:39:02,800 That just poses something close to the link that I looked at. 511 00:39:02,800 --> 00:39:07,240 And I think that explains it very well. 512 00:39:07,240 --> 00:39:08,240 Okay. 513 00:39:08,240 --> 00:39:11,280 But you're 100% right. 514 00:39:11,280 --> 00:39:15,320 I put it in the chat. 515 00:39:15,320 --> 00:39:19,480 So there's different words for these. 516 00:39:19,480 --> 00:39:24,760 And so that's actually one of the challenges you'll run into here in the psychology literature 517 00:39:24,760 --> 00:39:30,760 is they're kind of always using different terminology for these. 518 00:39:30,760 --> 00:39:37,400 But I just tried to standardize it the best I could. 519 00:39:37,400 --> 00:39:39,960 But you're 100% right, Jerry. 520 00:39:39,960 --> 00:39:40,960 And that... 521 00:39:40,960 --> 00:39:48,240 Well, if you look at what Candice linked you put in the chat, it gives you the both sides 522 00:39:48,240 --> 00:39:49,240 of the scale. 523 00:39:49,240 --> 00:39:50,240 Spot on. 524 00:39:50,240 --> 00:39:53,960 So we clearly raised... 525 00:39:53,960 --> 00:40:00,360 So this is what's awesome about presenting this research to you is we get to find out 526 00:40:00,360 --> 00:40:04,880 really quickly what points need more clarification. 527 00:40:04,880 --> 00:40:08,880 And that's what's awesome about working with you all. 528 00:40:08,880 --> 00:40:11,600 So this is not just... 529 00:40:11,600 --> 00:40:14,160 This is a team effort. 530 00:40:14,160 --> 00:40:22,960 In my book, this is team research because your feedback is, in my opinion, part of the 531 00:40:22,960 --> 00:40:24,640 most critical step. 532 00:40:24,640 --> 00:40:32,320 So here, I presented an idea to you and then you bounced back that, hey, we need to much 533 00:40:32,320 --> 00:40:37,080 clearly define these personality traits. 534 00:40:37,080 --> 00:40:43,520 The scale seems like a good idea, but we need to clearly define both ends of the scale. 535 00:40:43,520 --> 00:40:44,520 Awesome. 536 00:40:44,520 --> 00:40:53,200 So that's essentially all the points you raised have already improved the research. 537 00:40:53,200 --> 00:41:01,720 So just by coming and sharing your thoughts, you, in my opinion, helped advance this research 538 00:41:01,720 --> 00:41:05,920 question, which I feel is advancing cannabis science. 539 00:41:05,920 --> 00:41:07,320 So that's what I mean by... 540 00:41:07,320 --> 00:41:13,840 When I say, by you coming, I think you're helping advance cannabis science. 541 00:41:13,840 --> 00:41:21,160 As I said, it may be a snowflake, but snowflakes add up. 542 00:41:21,160 --> 00:41:30,600 So Keegan, if this really is of interest, there are established scales for this. 543 00:41:30,600 --> 00:41:36,300 In particular, we're very familiar with one called Brunel Mood that we have used with 544 00:41:36,300 --> 00:41:38,640 cannabis data. 545 00:41:38,640 --> 00:41:46,720 It allows respondents to respond to a series of words that roll up into key moods. 546 00:41:46,720 --> 00:41:55,040 We've looked, or we can continue to look, at how the acute cannabis response changes 547 00:41:55,040 --> 00:41:56,220 your mood. 548 00:41:56,220 --> 00:41:58,880 But one needs to select a scale. 549 00:41:58,880 --> 00:42:00,760 Scale means a survey. 550 00:42:00,760 --> 00:42:10,080 So for example, if folks want to go down this pathway, I would advocate looking very seriously 551 00:42:10,080 --> 00:42:14,000 at established scales like Brunel Mood. 552 00:42:14,000 --> 00:42:17,240 It's quite robust. 553 00:42:17,240 --> 00:42:18,240 I love it, John. 554 00:42:18,240 --> 00:42:23,080 And I love that you've already researched this topic. 555 00:42:23,080 --> 00:42:28,800 I could present on this at a future meeting because, yeah, we have a good number of years 556 00:42:28,800 --> 00:42:32,440 working together with Jeff Tarrant and Eugene on this. 557 00:42:32,440 --> 00:42:38,280 So if this is relevant, let's, in a future meeting, I'll walk you through some of our 558 00:42:38,280 --> 00:42:40,920 Brunel's data or Brunel's interpretations. 559 00:42:40,920 --> 00:42:44,320 Well, we may have an awesome platform for you. 560 00:42:44,320 --> 00:42:53,360 So hold tight because we may actually have a good time for you to present your... 561 00:42:53,360 --> 00:42:58,240 John, could you put that scale in the chat so we get the correct spelling? 562 00:42:58,240 --> 00:43:03,480 You know, Jerry, I don't do that kind of stuff very well. 563 00:43:03,480 --> 00:43:06,800 Can I just suggest that you look up Brunel Mood? 564 00:43:06,800 --> 00:43:07,800 I can spell it. 565 00:43:07,800 --> 00:43:09,800 Spell it for me, yeah. 566 00:43:09,800 --> 00:43:11,800 Yeah, B-R-U-N-E-L. 567 00:43:11,800 --> 00:43:12,800 Yep. 568 00:43:12,800 --> 00:43:15,440 You'll have no problem finding it. 569 00:43:15,440 --> 00:43:17,120 Okay, Brunel Mood. 570 00:43:17,120 --> 00:43:23,360 Brunel Mood or Brums, it's sometimes called Brunel Mood scale, B-R-U-M-S. 571 00:43:23,360 --> 00:43:24,640 We'll find it relevant for it. 572 00:43:24,640 --> 00:43:25,640 Thank you. 573 00:43:25,640 --> 00:43:26,640 Very widely available. 574 00:43:26,640 --> 00:43:36,680 And John, you've now helped our literature review because with my literature review failed 575 00:43:36,680 --> 00:43:40,720 to unearth, you effortlessly provided. 576 00:43:40,720 --> 00:43:46,840 So now here I'm shooting in the dark and all of a sudden you just now pointed us to, oh, 577 00:43:46,840 --> 00:43:49,480 you know, this is the metric you need to use. 578 00:43:49,480 --> 00:43:55,040 And so this is sort of my point is, you know, instead of me now going off on a wild goose 579 00:43:55,040 --> 00:44:05,200 chase for months on end, which I see happen all the time, like at universities and whatnot, 580 00:44:05,200 --> 00:44:06,200 right? 581 00:44:06,200 --> 00:44:09,720 I could have gone on a six month goose chase. 582 00:44:09,720 --> 00:44:13,560 And then I would have then presented my work to John and he said, oh, have you tried the 583 00:44:13,560 --> 00:44:16,320 Brunel Mood test scale? 584 00:44:16,320 --> 00:44:19,720 And then I would have been like, oh, well, no, back to the drawing board. 585 00:44:19,720 --> 00:44:26,680 And so basically now instead of wasting all that time, I thought of this and then one 586 00:44:26,680 --> 00:44:32,320 day later get to bounce it off of John, gets bounced back to my court. 587 00:44:32,320 --> 00:44:43,480 So the idea is we can just do this rapid, rapid research that when you spend a bit of 588 00:44:43,480 --> 00:44:54,920 time around, you know, university professors, one time a university professor specifically told me like, you know, things move slowly here. 589 00:44:54,920 --> 00:44:56,840 You just have to get used to that. 590 00:44:56,840 --> 00:45:01,720 So I'm sort of a go, go, go type of person. 591 00:45:01,720 --> 00:45:07,480 And so whether research should be done that way or not, that's an opinion. 592 00:45:07,480 --> 00:45:12,560 But I think in this open manner, we can do research quickly. 593 00:45:12,560 --> 00:45:21,360 And at the Brunel Mood, and I just put the link in the chat of an assessment, and it 594 00:45:21,360 --> 00:45:31,440 may be too, it may be so detailed as to make any result that we would come up with be insignificant. 595 00:45:31,440 --> 00:45:40,560 I mean, it has 24 different mood characteristics everywhere from panicky to uncertain. 596 00:45:40,560 --> 00:45:43,080 We want to add rigor to our analysis, Jerry. 597 00:45:43,080 --> 00:45:44,400 So this could be a good way to do it. 598 00:45:44,400 --> 00:45:49,360 So Jerry, we have we use it in the acute setting. 599 00:45:49,360 --> 00:45:54,240 So we ask folks before they get high to fill this out. 600 00:45:54,240 --> 00:45:57,680 You fill it out really fast because you're you're simple. 601 00:45:57,680 --> 00:46:04,680 It's like a button click for does the word apply or how much the word applies at that moment. 602 00:46:04,680 --> 00:46:09,280 I think what what Keegan's looking for is personality type in that mood. 603 00:46:09,280 --> 00:46:15,040 Yeah, and that so brooms is an example of mood. 604 00:46:15,040 --> 00:46:21,280 It would be if we're doing personality, we would probably find another scale or need 605 00:46:21,280 --> 00:46:22,280 to. 606 00:46:22,280 --> 00:46:27,880 I'm just pointing out that we have experience with the mood, the acute mood change using 607 00:46:27,880 --> 00:46:28,880 Brunel. 608 00:46:28,880 --> 00:46:30,880 OK, it's actually interesting. 609 00:46:30,880 --> 00:46:37,040 The real point I was trying to make is not binary, but this Langford scale or whatever 610 00:46:37,040 --> 00:46:41,360 is being important because I think I hope that's clear. 611 00:46:41,360 --> 00:46:45,880 You know, the binary assessment is such a cool way of doing things. 612 00:46:45,880 --> 00:46:54,480 But I think we're we lose, unfortunately, detail that way or the ability to to find 613 00:46:54,480 --> 00:46:57,200 correlates or factor analysis or whatever. 614 00:46:57,200 --> 00:47:03,720 If we're in binary mode is what Terrence was discussing with us this past week. 615 00:47:03,720 --> 00:47:11,160 I like your mood analysis and that actually will be relevant to a quick a quick something 616 00:47:11,160 --> 00:47:13,200 quick I want to show you today. 617 00:47:13,200 --> 00:47:18,580 But the mood analysis could also be done to see if that's relevant. 618 00:47:18,580 --> 00:47:24,920 But the other point I wanted to raise is the economist would come in and say, well, well, 619 00:47:24,920 --> 00:47:30,480 well, well, well, well, you know, you can't just be asking people about, you know, their 620 00:47:30,480 --> 00:47:33,560 personality traits here. 621 00:47:33,560 --> 00:47:39,960 You know, what if somebody I wanted to mislead you for for some reason or another. 622 00:47:39,960 --> 00:47:47,680 So or what if somebody just doesn't have time to take a personality test or this or that? 623 00:47:47,680 --> 00:47:54,960 Well, this was a novel idea that other people have had, but we can now use it here in the 624 00:47:54,960 --> 00:47:56,600 cannabis industry. 625 00:47:56,600 --> 00:48:05,960 So I found a handful of examples of people doing this, as I said, on Twitter posts and 626 00:48:05,960 --> 00:48:09,080 I saw eBay things. 627 00:48:09,080 --> 00:48:14,800 So there's many uses for this, but no one's done it in the cannabis industry yet. 628 00:48:14,800 --> 00:48:18,200 So I think we could maybe take a stab at it. 629 00:48:18,200 --> 00:48:26,980 And OK, the idea is we've collected the consumer purchases and the lab results. 630 00:48:26,980 --> 00:48:32,440 We also have the reviews for these purchases. 631 00:48:32,440 --> 00:48:40,880 So now what we could do is we could use natural language processing. 632 00:48:40,880 --> 00:48:50,200 Look at somebody's review and try to scale it on zero to one for each of the five personality 633 00:48:50,200 --> 00:48:51,200 traits. 634 00:48:51,200 --> 00:48:54,720 And why do I think that we can do this? 635 00:48:54,720 --> 00:49:02,720 Well, there's been psychologists that have hypothesized that personality traits would 636 00:49:02,720 --> 00:49:14,520 probably show up in language and specifically that perhaps, you know, really important personality 637 00:49:14,520 --> 00:49:20,320 traits may often get boiled down to single words. 638 00:49:20,320 --> 00:49:23,440 So it may not be the case. 639 00:49:23,440 --> 00:49:25,200 It's not tested. 640 00:49:25,200 --> 00:49:27,800 This is just a hypothesis. 641 00:49:27,800 --> 00:49:29,880 Or it's not a theory yet. 642 00:49:29,880 --> 00:49:31,160 It's just a hypothesis. 643 00:49:31,160 --> 00:49:33,080 We need to rigorously test it. 644 00:49:33,080 --> 00:49:40,880 But, I mean, what if it's the case that certain words are only used by really extroverted 645 00:49:40,880 --> 00:49:49,880 people or maybe people who are really high in the degree of openness may use the word 646 00:49:49,880 --> 00:49:56,360 open a lot or, you know, who knows what this is? 647 00:49:56,360 --> 00:50:02,800 The idea is, you know, we could just read through a bunch of people's reviews and do 648 00:50:02,800 --> 00:50:04,960 the best we could at it. 649 00:50:04,960 --> 00:50:13,320 But we could, the idea is, use natural language processing and spend much, much more time 650 00:50:13,320 --> 00:50:14,320 on it. 651 00:50:14,320 --> 00:50:20,240 So instead of reading through each sentence, you know, really, really carefully, you know, 652 00:50:20,240 --> 00:50:28,480 we can just let the computer read through these sentences at just lightning fast rate, 653 00:50:28,480 --> 00:50:33,440 perform logical operations at lightning fast rates. 654 00:50:33,440 --> 00:50:40,640 So all it's doing is what a human would do, which would be read the review. 655 00:50:40,640 --> 00:50:41,880 Okay. 656 00:50:41,880 --> 00:50:48,760 You know, we could train this on some data so we can actually say, okay, here's a data 657 00:50:48,760 --> 00:50:52,360 set where someone's taken a psychology test. 658 00:50:52,360 --> 00:50:55,600 Here are their actual traits. 659 00:50:55,600 --> 00:51:01,600 And then we can, you know, see how well we can predict those with the language they use. 660 00:51:01,600 --> 00:51:05,280 So that's the idea. 661 00:51:05,280 --> 00:51:13,200 And the platform for this is, so I'll just show you an example. 662 00:51:13,200 --> 00:51:19,520 And I was thinking about, I was going to float it by the group here before, you know, formally 663 00:51:19,520 --> 00:51:25,720 booking it, but I was thinking, okay, Saturday morning is maybe not the best time for people. 664 00:51:25,720 --> 00:51:33,320 So I was thinking about moving Saturday morning statistics to Thursday. 665 00:51:33,320 --> 00:51:39,200 And I was just going to see if anyone's thoughts about this and if this time may work for you. 666 00:51:39,200 --> 00:51:48,360 But maybe, you know, Thursday, I should have said from 420 to about 530 Eastern Standard 667 00:51:48,360 --> 00:51:49,360 Time. 668 00:51:49,360 --> 00:51:53,640 So the first 10 minutes or so will be pretty casual. 669 00:51:53,640 --> 00:51:59,200 And then we'll do about an hour of, you know, rigorous analytics. 670 00:51:59,200 --> 00:52:04,360 So that way, you know, the Wednesday group, we're just kind of having this light discussion. 671 00:52:04,360 --> 00:52:12,960 And then, you know, next Thursday, we'll actually get into the nitty gritty of this and actually 672 00:52:12,960 --> 00:52:19,720 go through all of the, right, because, you know, a criticism that's been levied is, okay, 673 00:52:19,720 --> 00:52:21,720 we don't, we're not rigorous enough. 674 00:52:21,720 --> 00:52:25,240 We don't do enough training and testing of our models. 675 00:52:25,240 --> 00:52:34,200 And we may be providing misleading information because we're giving the illusion that we're 676 00:52:34,200 --> 00:52:42,920 doing statistics, but the statistics aren't robust and people may be drawing incorrect 677 00:52:42,920 --> 00:52:45,720 inferences from those statistics. 678 00:52:45,720 --> 00:52:51,240 So the idea is we'll do a really, really robust session here. 679 00:52:51,240 --> 00:53:02,600 So that way, you know, the Wednesday group, pretty casual and then hardcore on Thursday 680 00:53:02,600 --> 00:53:11,360 where rigorous statistics, you get the results, we actually will have some faith in our results, 681 00:53:11,360 --> 00:53:12,360 right? 682 00:53:12,360 --> 00:53:21,520 So instead of just saying, oh, this is a mock or this is a proof of concept, no, like where 683 00:53:21,520 --> 00:53:31,160 this is a rigorous analysis, we've tested, we've trained, we're confident, as confident 684 00:53:31,160 --> 00:53:33,120 as we can be. 685 00:53:33,120 --> 00:53:36,880 And you know, this is advanced cannabis analytics, right? 686 00:53:36,880 --> 00:53:44,080 We're taking the most rigorous statistics out there, applying it rigorously to the cannabis 687 00:53:44,080 --> 00:53:53,000 field in an open manner that anyone can participate in, anyone can contribute. 688 00:53:53,000 --> 00:53:55,400 So that's the idea here. 689 00:53:55,400 --> 00:54:07,040 My question about language is different people from different social groups use language 690 00:54:07,040 --> 00:54:17,200 differently, especially non-English language, you know, first like either non-English speakers 691 00:54:17,200 --> 00:54:21,040 or people whose English is not their first language. 692 00:54:21,040 --> 00:54:27,840 So you raise this point, Jerry, and I don't know if it's a good or bad thing, but the 693 00:54:27,840 --> 00:54:35,520 first thing that comes to my mind is, and this is from a statistical point of view, 694 00:54:35,520 --> 00:54:44,840 a statistics professor once told me he looks for variability in his data, because if there's 695 00:54:44,840 --> 00:54:49,640 variability, you'll be able to do good statistics with it, right? 696 00:54:49,640 --> 00:54:53,200 Statistics is the science of uncertainty. 697 00:54:53,200 --> 00:54:59,240 So the more uncertainty you have, the better statistics is a tool for that job. 698 00:54:59,240 --> 00:55:03,900 If there's not much uncertainty, you can just go to math. 699 00:55:03,900 --> 00:55:11,440 So I think we could actually use that essentially to our advantage in that, okay, that may very 700 00:55:11,440 --> 00:55:17,680 well be the case, but introverts and extroverts may still use language differently. 701 00:55:17,680 --> 00:55:24,880 So maybe English is somebody's second language, but maybe that may even make them even more 702 00:55:24,880 --> 00:55:29,060 predisposed to using certain words. 703 00:55:29,060 --> 00:55:36,720 So it could even be the case that we could predict personality traits even better if 704 00:55:36,720 --> 00:55:39,080 English is somebody's second language. 705 00:55:39,080 --> 00:55:48,720 May not be the case, but I think it does add more variability, but I think potentially 706 00:55:48,720 --> 00:55:52,920 could use it to our advantage, but it's definitely a good consideration. 707 00:55:52,920 --> 00:55:57,400 Can I make a couple of points? 708 00:55:57,400 --> 00:55:59,680 And I think there's an elephant in this room. 709 00:55:59,680 --> 00:56:02,920 I mean, it's an interesting idea. 710 00:56:02,920 --> 00:56:13,640 What I might propose is, so you brought up earlier the chemotyping dichotomy or chemotyping 711 00:56:13,640 --> 00:56:19,420 classification that I'm certainly advocating that folks adopt and ride. 712 00:56:19,420 --> 00:56:27,620 So let's just say, for example, that's one of our principal discriminators. 713 00:56:27,620 --> 00:56:33,800 So the chemotype of the botanical that's reported on, let's say that's point one. 714 00:56:33,800 --> 00:56:39,720 So you want to see if you can use natural language processing or some sort of computer 715 00:56:39,720 --> 00:56:42,920 algorithm to parse reviews. 716 00:56:42,920 --> 00:56:51,400 Maybe the parameter there is how active or passively written the text is. 717 00:56:51,400 --> 00:56:58,000 And what I mean by active is I used this bud to get high as active. 718 00:56:58,000 --> 00:57:04,880 I would say this product made me sleepy as passive. 719 00:57:04,880 --> 00:57:09,800 These are pretty important semantic distinctions. 720 00:57:09,800 --> 00:57:16,480 And it may be quite easy to distinguish between an active review and a passive review. 721 00:57:16,480 --> 00:57:18,120 So you score it that way. 722 00:57:18,120 --> 00:57:23,240 That's all fine and dandy, but it hangs when you're done. 723 00:57:23,240 --> 00:57:26,880 You've got to tie it to something for it to be relevant. 724 00:57:26,880 --> 00:57:28,400 And that's the dilemma I have. 725 00:57:28,400 --> 00:57:29,400 So what? 726 00:57:29,400 --> 00:57:35,400 You come up with at the end two thirds of people like this chemotype who write active 727 00:57:35,400 --> 00:57:38,880 reviews and one third like the other ones. 728 00:57:38,880 --> 00:57:39,880 Then what? 729 00:57:39,880 --> 00:57:42,920 I mean, where do you go from there? 730 00:57:42,920 --> 00:57:46,440 That's the dilemma that I see. 731 00:57:46,440 --> 00:57:49,080 It's just a pretty thing to hang out there. 732 00:57:49,080 --> 00:57:52,000 It doesn't tie to anything. 733 00:57:52,000 --> 00:57:56,600 And so I would encourage us to be thinking about, so what do you tie this kind of analysis 734 00:57:56,600 --> 00:57:58,800 or this kind of report to? 735 00:57:58,800 --> 00:58:02,300 The way you raised many good points. 736 00:58:02,300 --> 00:58:10,120 So to address them in reverse order, what I would tie it to is predicting our hat. 737 00:58:10,120 --> 00:58:24,960 So when you build this model, we often build coming from a learning statistics from a university, 738 00:58:24,960 --> 00:58:30,920 people are often interested in estimating beta because they just want to know the relationship 739 00:58:30,920 --> 00:58:39,160 between say openness and somebody's average ratio. 740 00:58:39,160 --> 00:58:44,800 But in the business world, you're more interested in prediction. 741 00:58:44,800 --> 00:58:49,660 So in the business world, you're more interested in predicting our hat. 742 00:58:49,660 --> 00:58:56,100 So given O, C, E, A, and N, what would be our hat? 743 00:58:56,100 --> 00:59:00,160 And so the idea is I think this could benefit a consumer. 744 00:59:00,160 --> 00:59:07,880 So most people generally know their, well actually, well I don't off the top of my head, 745 00:59:07,880 --> 00:59:10,040 so that may not be a correct statement. 746 00:59:10,040 --> 00:59:17,760 But the idea is as a consumer, you could basically take a personality test and one, you'd get 747 00:59:17,760 --> 00:59:19,040 your personality. 748 00:59:19,040 --> 00:59:24,980 And then two, you could say, oh, on average, people with your personality prefer this type 749 00:59:24,980 --> 00:59:26,620 of string. 750 00:59:26,620 --> 00:59:34,460 So that way, before you've even maybe before you've even tried cannabis for the first time, 751 00:59:34,460 --> 00:59:42,120 you now know what people with your personality, what type of strains they may gravitate towards. 752 00:59:42,120 --> 00:59:45,120 It may be, there may be absolutely no difference. 753 00:59:45,120 --> 00:59:49,100 Maybe everybody, maybe personality has no effect. 754 00:59:49,100 --> 00:59:52,920 So remember all these betas, these could all be zero. 755 00:59:52,920 --> 00:59:55,120 So that's the idea is we want to test to see. 756 00:59:55,120 --> 01:00:01,600 I would say that people who market and produce and sell cannabis would want this information. 757 01:00:01,600 --> 01:00:09,320 And that's the idea is, oh, these psychology traits, we're saying, oh, they get incorporated 758 01:00:09,320 --> 01:00:11,200 into language. 759 01:00:11,200 --> 01:00:15,960 Well marketers are all about their marketing type. 760 01:00:15,960 --> 01:00:24,320 So it's like, oh, if you've got a strain that you think extroverts will love, maybe 761 01:00:24,320 --> 01:00:27,600 you can somehow market that to extroverts. 762 01:00:27,600 --> 01:00:32,480 Say, oh, you'll love this as a party or something like that. 763 01:00:32,480 --> 01:00:37,360 I'm just, once I said, I'm just kind of stabbing there. 764 01:00:37,360 --> 01:00:42,440 Psychologies I need to do a lot of reading up on. 765 01:00:42,440 --> 01:00:48,520 But the idea is maybe like Jerry said, maybe marketers could use this in tailoring products 766 01:00:48,520 --> 01:00:51,960 to certain personality types. 767 01:00:51,960 --> 01:01:01,600 And then as I said, maybe consumers could use it to maybe make an educated guess about 768 01:01:01,600 --> 01:01:09,480 what product they may like. 769 01:01:09,480 --> 01:01:15,120 So that's the main idea in a nutshell. 770 01:01:15,120 --> 01:01:18,520 As I said, a lot of the code is already written. 771 01:01:18,520 --> 01:01:23,040 But your first point was how do we actually assign the traits? 772 01:01:23,040 --> 01:01:28,480 And that's what I want to end up spending most of the time on Thursday talking about. 773 01:01:28,480 --> 01:01:31,920 Because that's sort of where all the fun comes. 774 01:01:31,920 --> 01:01:36,280 Because I said, I'm not the first one who's done this, but basically kind of like what 775 01:01:36,280 --> 01:01:44,520 John's saying is, do people use passive or active sentences, so on and so forth? 776 01:01:44,520 --> 01:01:51,440 And do those choices help predict personality? 777 01:01:51,440 --> 01:01:54,840 And then there's a couple training datasets we can use. 778 01:01:54,840 --> 01:02:01,840 So there's basically a training dataset where someone's written an essay and then they've 779 01:02:01,840 --> 01:02:04,320 also taken a personality test. 780 01:02:04,320 --> 01:02:09,520 So we can basically build a training, so we can train our model on the training dataset 781 01:02:09,520 --> 01:02:14,080 and then we'll use it on our actual dataset. 782 01:02:14,080 --> 01:02:18,520 So that's how we'll go about using natural language processing. 783 01:02:18,520 --> 01:02:22,880 But that could almost use a whole meetup of its own. 784 01:02:22,880 --> 01:02:27,320 So instead of Saturday, it'll be next Thursday. 785 01:02:27,320 --> 01:02:30,560 And I think this one's going to be cool. 786 01:02:30,560 --> 01:02:38,080 So today presented the methodology and then next Thursday we'll actually estimate it. 787 01:02:38,080 --> 01:02:44,200 So now I think I'm just kind of just going on and I want to make sure everybody can get 788 01:02:44,200 --> 01:02:45,520 on with their day. 789 01:02:45,520 --> 01:02:50,920 So any last thoughts, comments, questions before we call it a day here? 790 01:02:50,920 --> 01:03:02,520 For NLP processing, is it difficult to get personality traits of introverted persons 791 01:03:02,520 --> 01:03:14,040 from language because they mostly don't review products or they don't participate in social 792 01:03:14,040 --> 01:03:20,400 media and I guess it's difficult to get personality of introverted persons. 793 01:03:20,400 --> 01:03:22,520 You raised an awesome point. 794 01:03:22,520 --> 01:03:31,040 And basically what you raised is, is there a systemic difference between people who are 795 01:03:31,040 --> 01:03:33,960 actually going about answering reviews? 796 01:03:33,960 --> 01:03:37,100 And this easily could be the case. 797 01:03:37,100 --> 01:03:45,880 So I talked about this where like the positivity bias, so maybe people who had a positive experience 798 01:03:45,880 --> 01:03:47,680 are more likely to leave reviews. 799 01:03:47,680 --> 01:03:53,680 And as you mentioned, maybe people with certain personality traits are more likely to leave 800 01:03:53,680 --> 01:03:55,640 reviews. 801 01:03:55,640 --> 01:04:01,360 I think that should be okay as long as our training data isn't biased. 802 01:04:01,360 --> 01:04:06,880 So our training data should be about equal equal. 803 01:04:06,880 --> 01:04:10,760 So actually I don't even know if that necessarily matters. 804 01:04:10,760 --> 01:04:16,840 I don't know if you need an equidistribution of observations. 805 01:04:16,840 --> 01:04:20,400 So I think it's an awesome point. 806 01:04:20,400 --> 01:04:22,600 I'll need to think about this. 807 01:04:22,600 --> 01:04:28,720 I want to think it's, I think we can work with this. 808 01:04:28,720 --> 01:04:31,480 It'll just be our observed data. 809 01:04:31,480 --> 01:04:35,160 There just may be more extroverts that we observe. 810 01:04:35,160 --> 01:04:36,660 That should be okay. 811 01:04:36,660 --> 01:04:39,000 But I think what you've raised is an excellent point. 812 01:04:39,000 --> 01:04:42,500 And it kind of hits on sampling bias. 813 01:04:42,500 --> 01:04:45,380 So this is a critical bias you've got to watch out for. 814 01:04:45,380 --> 01:04:47,380 So thank you. 815 01:04:47,380 --> 01:04:51,720 So thank you for bringing this point up. 816 01:04:51,720 --> 01:04:59,680 I think if one goes down this pathway and so at the end the results that you get, you 817 01:04:59,680 --> 01:05:08,560 get a predictive model or you get a frequency count or an incidence count or something. 818 01:05:08,560 --> 01:05:14,520 So after you've done all your machinations, you come up with okay two-thirds fall in this 819 01:05:14,520 --> 01:05:16,920 group, one-third fall in this group. 820 01:05:16,920 --> 01:05:17,920 So what? 821 01:05:17,920 --> 01:05:22,400 You have to, I believe, then pair this with something else. 822 01:05:22,400 --> 01:05:29,560 And that might be actually some kind of interventional study or interventional trial where you then 823 01:05:29,560 --> 01:05:35,560 have, it sets the hypothesis for the group and then you have to go out and actually do 824 01:05:35,560 --> 01:05:39,560 it with people and see how well it matches. 825 01:05:39,560 --> 01:05:40,940 Maybe that's the approach. 826 01:05:40,940 --> 01:05:47,760 So what you're building initially is a hypothesis with the motivation to then go fund and do 827 01:05:47,760 --> 01:05:50,820 some kind of interventional study. 828 01:05:50,820 --> 01:05:56,760 Maybe that's its utility, but I'm concerned if you just do it in isolation, it hangs and 829 01:05:56,760 --> 01:06:01,320 it's just another shiny pretty picture on the wall. 830 01:06:01,320 --> 01:06:04,880 In John, you raised two critical points. 831 01:06:04,880 --> 01:06:14,880 The first being is even this beta-pinene-d-lemonene ratio, the key metric we should look at or 832 01:06:14,880 --> 01:06:17,580 maybe we want to look at product type, et cetera. 833 01:06:17,580 --> 01:06:24,560 So our outcome variable we may want to think about, so maybe extroverts are more likely 834 01:06:24,560 --> 01:06:26,160 to try, so does. 835 01:06:26,160 --> 01:06:29,320 So there's many research questions we could ask there. 836 01:06:29,320 --> 01:06:34,880 The second more important point you raised is, is this essentially just a hypothesis? 837 01:06:34,880 --> 01:06:37,360 And my answer is absolutely. 838 01:06:37,360 --> 01:06:45,160 It's essentially the value we're adding here is I think we're bringing forth or informalizing, 839 01:06:45,160 --> 01:06:48,080 thankfully, it's not a novel hypothesis. 840 01:06:48,080 --> 01:06:54,000 At least we're hopefully stating it slightly more formally. 841 01:06:54,000 --> 01:06:58,680 But this is, like I said, a hypothesis that's been on my mind, so I just kind of wanted 842 01:06:58,680 --> 01:07:01,080 to formally state it. 843 01:07:01,080 --> 01:07:06,440 And then exactly, it's sort of that picture of the law of large numbers that we saw. 844 01:07:06,440 --> 01:07:14,360 We're just t equals one, so we're just showing how it could be done with this ad-hoc dataset 845 01:07:14,360 --> 01:07:18,520 that we happen to have in the public domain. 846 01:07:18,520 --> 01:07:26,920 But if you then go and do a clinical study or do a study with some of your own data, 847 01:07:26,920 --> 01:07:31,680 then that'll be t equals two, t equals three, so on and so forth. 848 01:07:31,680 --> 01:07:40,160 And so the more and more you test this, the more likely the tests are to converge on the 849 01:07:40,160 --> 01:07:41,840 actual effect. 850 01:07:41,840 --> 01:07:47,440 So do I think we're going to actually measure the effect, the true effect, our first time 851 01:07:47,440 --> 01:07:50,680 we study this, the first model we estimate? 852 01:07:50,680 --> 01:07:51,760 Unlikely. 853 01:07:51,760 --> 01:07:56,760 But hopefully the idea is we can get a lot of great minds thinking about this, and then 854 01:07:56,760 --> 01:08:02,120 many, many people can estimate this model, test out this relationship, and then in the 855 01:08:02,120 --> 01:08:09,820 long run, hopefully we can figure out if there's any rhyme or reason to personality in cannabis. 856 01:08:09,820 --> 01:08:11,880 Is there anything to it? 857 01:08:11,880 --> 01:08:13,880 I think it's interesting to think about. 858 01:08:13,880 --> 01:08:17,720 So I'll let you all chew on this until next week. 859 01:08:17,720 --> 01:08:23,040 Well, we've gone over, so I'll let you get out of here and enjoy your day. 860 01:08:23,040 --> 01:08:24,040 Just want you to thank you. 861 01:08:24,040 --> 01:08:25,040 Thank you all for coming. 862 01:08:25,040 --> 01:08:26,040 Very interesting again, Keenan. 863 01:08:26,040 --> 01:08:27,040 See you later. 864 01:08:27,040 --> 01:08:28,040 Good discussion. 865 01:08:28,040 --> 01:08:29,040 Thanks. 866 01:08:29,040 --> 01:08:30,040 Thank you, Keegan. 867 01:08:30,040 --> 01:08:31,040 Thank you, everyone. 868 01:08:31,040 --> 01:08:32,040 Bye now. 869 01:08:32,040 --> 01:08:37,800 Bye now.