1 00:00:00,000 --> 00:00:11,240 Welcome to the Canvas Data Science Meetup group. 2 00:00:11,240 --> 00:00:13,040 Happy to have you all here. 3 00:00:13,040 --> 00:00:17,600 It could be the best Canvas Data Science Meetup yet. 4 00:00:17,600 --> 00:00:21,200 There's some real cool things that we'll cover today. 5 00:00:21,200 --> 00:00:24,920 I think you'll be in for a treat. 6 00:00:25,240 --> 00:00:27,600 Without further ado, 7 00:00:27,600 --> 00:00:33,880 how about I jump into it and then we can talk about the data and statistics 8 00:00:33,880 --> 00:00:37,880 in Canvas Insights as we go. 9 00:00:37,880 --> 00:00:49,720 So for those of you who are new to the group, 10 00:00:49,720 --> 00:00:56,320 we started last February and we've been collecting data 11 00:00:56,320 --> 00:00:58,720 and we've been collecting data from here. 12 00:00:58,720 --> 00:01:00,320 We've hit so many cities. 13 00:01:00,320 --> 00:01:03,560 We've hit Oregon, 14 00:01:03,560 --> 00:01:06,680 we've hit Colorado, Washington, 15 00:01:06,680 --> 00:01:10,520 we've been walking a lot in Massachusetts, 16 00:01:10,520 --> 00:01:15,280 and really just looking at states all over the country. 17 00:01:15,280 --> 00:01:22,480 And so the reason we're doing this is we're trying to essentially build 18 00:01:22,480 --> 00:01:30,160 nice collect systems so that way people can collect data from themselves. 19 00:01:30,160 --> 00:01:37,520 So here I've just started an ongoing list of all the states, 20 00:01:37,520 --> 00:01:43,360 whether you or not they have cannabis together or not, 21 00:01:43,360 --> 00:01:49,200 and then we've done without collecting the data. 22 00:01:49,200 --> 00:01:54,160 We've been collecting some of these and so some of these are near completion. 23 00:01:54,160 --> 00:01:58,480 So for example, your completion, 24 00:01:58,480 --> 00:02:05,480 Massachusetts, Connecticut, and really just five to a billion states. 25 00:02:05,480 --> 00:02:13,440 And so it will be exciting as we start to check off the box for each of these states. 26 00:02:13,440 --> 00:02:18,600 Washington also needs to start to move this towards completion, 27 00:02:18,600 --> 00:02:24,080 but still so much data we're going to do that are you ever finished. 28 00:02:24,080 --> 00:02:26,720 So this is a nice ongoing project. 29 00:02:26,720 --> 00:02:31,800 So if you want to put your hands on the scripts or contribute, 30 00:02:31,800 --> 00:02:36,360 definitely do the repository star. 31 00:02:36,360 --> 00:02:41,440 And then if you watch the repository, 32 00:02:41,440 --> 00:02:47,040 you can get notified whenever there are updates. 33 00:02:47,040 --> 00:02:50,800 So whenever I post the latest code, 34 00:02:50,800 --> 00:02:54,880 you will be able to make it to the meetup. 35 00:02:54,880 --> 00:03:03,360 Well, you can still get notified that the latest presentation was just uploaded. 36 00:03:03,360 --> 00:03:06,840 And in general, in from now on, 37 00:03:06,840 --> 00:03:12,200 we're going to try to upload the meetup one hour before the actual meetup. 38 00:03:12,200 --> 00:03:20,160 You can get a year in store, familiarize yourself with the script, 39 00:03:20,160 --> 00:03:25,480 or even get your hands on the data ahead of time. 40 00:03:25,480 --> 00:03:29,400 So that way you can get your hands on it. 41 00:03:29,400 --> 00:03:31,080 But it's not critical. 42 00:03:31,080 --> 00:03:33,400 This is here for perpetuity. 43 00:03:33,400 --> 00:03:36,800 So feel free to enjoy it. 44 00:03:36,800 --> 00:03:39,640 It's all licensed under the MIT license. 45 00:03:39,640 --> 00:03:44,680 So it includes the copyright and license, 46 00:03:44,680 --> 00:03:52,320 and you're off to the races to use the code however you de-fit. 47 00:03:52,320 --> 00:03:58,560 So that's just a quick introduction to what we're doing in our group. 48 00:03:58,560 --> 00:04:04,120 And so as I've said, we're in for a break today. 49 00:04:04,120 --> 00:04:11,440 So first thing first, start with the paper. 50 00:04:11,440 --> 00:04:20,040 So I've compiled all of the data that we've collected 51 00:04:20,040 --> 00:04:24,040 to the degree throughout the meetup. 52 00:04:24,040 --> 00:04:27,720 Do you have any other questions? 53 00:04:27,720 --> 00:04:32,720 Hey, I'm not trying to be a begging choose or whatever. 54 00:04:32,720 --> 00:04:36,720 Maybe somebody in the chat room, I can't access my chat. 55 00:04:36,720 --> 00:04:40,600 But the quality of maybe it's mine or your audio 56 00:04:40,600 --> 00:04:45,240 is kind of sporadic slash choppy. 57 00:04:45,240 --> 00:04:47,680 Oh, Graham, I've been trying to get in touch with you. 58 00:04:47,680 --> 00:04:51,640 Can I copy your LinkedIn format on your description? 59 00:04:51,640 --> 00:04:55,800 I don't really have a built up LinkedIn yet. 60 00:04:55,800 --> 00:04:59,800 I have one from prior to my disability. 61 00:04:59,800 --> 00:05:07,800 But you can totally get in touch with me with my LinkedIn. 62 00:05:07,800 --> 00:05:10,800 No, I can't if you have premium. 63 00:05:10,800 --> 00:05:13,800 So I'm going to leave. 64 00:05:13,800 --> 00:05:14,800 I'm going to leave. 65 00:05:14,800 --> 00:05:15,800 I'm going to leave. 66 00:05:15,800 --> 00:05:16,800 I'm going to leave. 67 00:05:16,800 --> 00:05:17,800 I'm going to leave. 68 00:05:17,800 --> 00:05:18,800 I'm going to leave. 69 00:05:18,800 --> 00:05:19,800 I'm going to leave. 70 00:05:19,800 --> 00:05:20,800 I'm going to leave. 71 00:05:20,800 --> 00:05:21,800 I'm going to leave. 72 00:05:21,800 --> 00:05:22,800 I'm going to leave. 73 00:05:22,800 --> 00:05:23,800 I'm going to leave. 74 00:05:23,800 --> 00:05:24,800 I'm going to leave. 75 00:05:24,800 --> 00:05:25,800 So I don't have premium. 76 00:05:25,800 --> 00:05:26,800 So I can't message you. 77 00:05:26,800 --> 00:05:27,800 So I tried connecting with you. 78 00:05:27,800 --> 00:05:28,800 I can't do that. 79 00:05:28,800 --> 00:05:36,160 So I have a year and a half of employment to a gap to explain. 80 00:05:36,160 --> 00:05:38,200 I can't be on camera right now. 81 00:05:38,200 --> 00:05:39,200 I'm pretty sick. 82 00:05:39,200 --> 00:05:40,200 Anyway, so great. 83 00:05:40,200 --> 00:05:41,200 Maybe I'll get in touch with you some other way. 84 00:05:41,200 --> 00:05:44,240 So anyway, Keegan, your audio is kind of jumping around. 85 00:05:44,240 --> 00:05:57,880 jumping around so not really maybe somebody else can comment maybe somebody 86 00:05:57,880 --> 00:06:01,800 else feels the same way I'll let them speak 87 00:06:01,800 --> 00:06:10,080 I can't reduce the amount of input but is it unbearable to continue or? 88 00:06:10,080 --> 00:06:16,040 not by my I mean I'm gonna let somebody else determine that but I feel like I 89 00:06:16,040 --> 00:06:20,360 can only hear every few of your words not every word so 90 00:06:20,360 --> 00:06:24,680 I think Keegan 91 00:06:24,680 --> 00:06:26,180 yes 92 00:06:26,180 --> 00:06:29,720 I think it's gonna need to be more of a visual 93 00:06:29,720 --> 00:06:36,120 you're gonna have to speak slower it's just because we have enough people in 94 00:06:36,120 --> 00:06:46,280 here that it may lag a little bit just because of the internet connection but 95 00:06:46,280 --> 00:06:57,560 um also I posted my email in the in call messages thank you thank you yeah and 96 00:06:57,560 --> 00:07:03,980 it's still Graham like the cracker but I would love to be in touch with people in 97 00:07:03,980 --> 00:07:10,760 Maryland that use medical cannabis thank you I was just um yeah because you have 98 00:07:10,760 --> 00:07:14,600 premium I wasn't able to connect so awesome thank you so much 99 00:07:14,600 --> 00:07:17,600 you're welcome 100 00:07:17,600 --> 00:07:24,880 okay am I still with you 101 00:07:24,880 --> 00:07:37,120 yeah okay well I'll try to do mostly people so today we are going to look at 102 00:07:37,120 --> 00:07:47,480 all the data we've collected through the the year so far so as you can see we've 103 00:07:47,480 --> 00:08:02,040 collected a lot of data points from many states so just a quick overview of of 104 00:08:02,040 --> 00:08:14,080 the goal of today so we've looked at the various states well today we're going to 105 00:08:14,080 --> 00:08:23,280 try to forecast for all these states so the ones we've at least collected data 106 00:08:23,280 --> 00:08:35,960 for so Illinois Massachusetts Washington Colorado and Oregon and we'll apply the 107 00:08:35,960 --> 00:08:43,520 ten commandments of forecasting so we're going to be forecasting sales we're 108 00:08:43,520 --> 00:08:52,600 doing this so that everybody has a nice transparent model for their own value 109 00:08:52,600 --> 00:08:56,080 and their own understanding of the industry so we're making this public 110 00:08:56,080 --> 00:09:06,240 knowledge we're acknowledging that our forecasts are imperfect so people should 111 00:09:06,240 --> 00:09:13,280 not put too much stake in our forecasts since we've done these fairly quickly 112 00:09:13,280 --> 00:09:29,320 our forecast horizon will be 2022 because much further than 2022 and we 113 00:09:29,320 --> 00:09:36,960 start to lose credibility and predictability and much shorter than that 114 00:09:36,960 --> 00:09:43,840 and it's not as valuable so we'll be predicting 2022 we're going to use an 115 00:09:43,840 --> 00:09:52,360 a theoretical model today so we'll just use total sales as well as just a few 116 00:09:52,360 --> 00:10:01,320 natural things like month effects we'll use the theoretical model just to keep it 117 00:10:01,320 --> 00:10:12,640 simple as a student as a reader Edward Tuft I always emphasize that you should 118 00:10:12,640 --> 00:10:21,720 put a lot of work into presenting your results in a beautiful informational way 119 00:10:21,720 --> 00:10:34,480 so visualizing data is key especially to explaining data to other people then 120 00:10:34,480 --> 00:10:43,840 we'll analyze our forecasts in the coolest thing is we can iterate on this 121 00:10:43,840 --> 00:10:53,720 so coming into 2022 we'll be able to judge all of our forecasts that we've 122 00:10:53,720 --> 00:11:00,960 made and then if our forecast model is really poor then we can think about ways 123 00:11:00,960 --> 00:11:08,400 to improve it so and then we'll go back to step one so this is sort of an 124 00:11:08,400 --> 00:11:18,360 iterative process that you can follow to forecast so without further ado let's do 125 00:11:18,360 --> 00:11:30,640 some forecasts so I've just read in a handful of useful packages here date 126 00:11:30,640 --> 00:11:36,720 youdle is a new one that we haven't used yet and so this will be using to 127 00:11:36,720 --> 00:11:44,280 manipulate time and so the other ones map plot live pandas PMD or EMA 128 00:11:44,280 --> 00:11:48,840 Seaborn we've touched on these before and then these are just a couple handy 129 00:11:48,840 --> 00:12:01,880 helper functions so first things first let's get our paws on the data that's 130 00:12:01,880 --> 00:12:28,440 not what we want so you know you can start to you know we can start to just 131 00:12:28,440 --> 00:12:33,680 do summary statistics right we've got total sales you know we're starting to 132 00:12:33,680 --> 00:12:38,440 look at retailers we've got the population we're even starting to get 133 00:12:38,440 --> 00:12:46,080 things like prices so we know we've got the data here now first things first we 134 00:12:46,080 --> 00:12:50,600 need to look at the data right because this is just a pile of numbers here so 135 00:12:50,600 --> 00:13:05,680 here I'm just going to add a time index simply to still getting new to VS codes 136 00:13:05,680 --> 00:13:12,400 hotkeys so bear with me I like the visual interface a little better 137 00:13:12,400 --> 00:13:19,840 sometimes for VS code but I'm still getting used to it so so bear with me 138 00:13:19,840 --> 00:13:24,760 but we've got some good visualizations coming up so let's just go ahead and and 139 00:13:24,760 --> 00:13:39,040 get to these so first things first let's just look at the total sales so this is 140 00:13:39,040 --> 00:13:46,120 the data that we've diligently collected starting in February and throughout the 141 00:13:46,120 --> 00:13:51,520 year I still have to add the Oklahoma data and then there are a few other 142 00:13:51,520 --> 00:13:56,880 states I think we could easily add well maybe not easily but we could try to add 143 00:13:56,880 --> 00:14:02,920 such as California I haven't tried to add Arizona yet so there's still a few 144 00:14:02,920 --> 00:14:11,200 more states to add but these are the ones that we've covered so Colorado of 145 00:14:11,200 --> 00:14:16,840 course at the top and so as you can see Colorado has quite a quite a head start 146 00:14:16,840 --> 00:14:23,360 so it'll be interesting to see if the other states you know get to the level 147 00:14:23,360 --> 00:14:34,000 of sales that Colorado is currently at Illinois has recently overtaken 148 00:14:34,000 --> 00:14:42,760 Washington for total sales so that's an interesting observation and then of 149 00:14:42,760 --> 00:14:52,320 course we've noted that Massachusetts is quite volatile so whereas you see 150 00:14:52,320 --> 00:15:06,320 Oregon Washington and Colorado increased sales in April and May of 2020 you see 151 00:15:06,320 --> 00:15:14,360 right despite the downward spike in sales to zero in Massachusetts as stores 152 00:15:14,360 --> 00:15:22,520 were closed and then they were just open for about a week in many so that's why 153 00:15:22,520 --> 00:15:28,160 sales are low during that period so drill interesting because Massachusetts 154 00:15:28,160 --> 00:15:35,800 is the only state that really closed stores so there is potentially you know 155 00:15:35,800 --> 00:15:42,400 differential analysis that could be done there to parse out various effects of 156 00:15:42,400 --> 00:15:52,640 cannabis consumption I mean there's a myriad of studies you could do so so to 157 00:15:52,640 --> 00:16:00,520 keep that in mind and then we just have limited sales data for Maryland and then 158 00:16:00,520 --> 00:16:06,360 here's Maine which has a small population which explains the small 159 00:16:06,360 --> 00:16:11,320 number of sales but it may even may have a disproportionately low number of 160 00:16:11,320 --> 00:16:18,720 sales and maybe because they're right next door in Massachusetts who knows 161 00:16:18,720 --> 00:16:30,080 one thing that Maine does have or prices just checking is the audio coming in to 162 00:16:30,080 --> 00:16:37,240 the point where you can follow along yes thank you so much awesome sorry I didn't 163 00:16:37,240 --> 00:16:41,440 do a good job of relaying back the appropriate feedback thank you so much 164 00:16:41,440 --> 00:16:48,280 thank you Heather just it's got to be otherwise what's the point of listening 165 00:16:48,280 --> 00:16:54,080 so okay anyway moving on we've got the price per gram in just a couple of 166 00:16:54,080 --> 00:17:01,840 states but this is a valuable data point and so it would be awesome to add this 167 00:17:01,840 --> 00:17:07,560 data point in more states we could add it in Washington it's just going to take 168 00:17:07,560 --> 00:17:16,640 a lot of work so be in be on the aisle for price data because valuable but what 169 00:17:16,640 --> 00:17:24,080 we can observe is you know Maine and Massachusetts have comparable prices 170 00:17:24,080 --> 00:17:32,320 with Maine originally being above Massachusetts and current day below 171 00:17:33,040 --> 00:17:43,960 interestingly Oregon is famous for their prices falling through the floor prices 172 00:17:43,960 --> 00:17:52,240 so it's really going to be tough to explain what's going on in Oregon so I 173 00:17:52,240 --> 00:17:59,360 don't have a ready explanation there may be things going on with you know 174 00:17:59,360 --> 00:18:03,560 black market cannabis or it just could be just a ton of supply there just 175 00:18:03,560 --> 00:18:09,880 couldn't just be it right licensing may be quite easy in Oregon so there may just 176 00:18:09,880 --> 00:18:18,120 be a ton of supply and price and prices have just gone down another thing is it 177 00:18:18,120 --> 00:18:24,880 could be a time effect so it could be that you know Oregon and Washington 178 00:18:24,880 --> 00:18:33,920 they've had cannabis for several for for many years now going on a decade for 179 00:18:33,920 --> 00:18:42,040 Washington and so you may just expect the market dynamics the you know the 180 00:18:42,040 --> 00:18:47,400 players that have operated in there they've found ways to cut prices right 181 00:18:47,400 --> 00:18:52,520 the producers have found ways to grow more efficiently people have found 182 00:18:52,520 --> 00:19:00,640 economies of steel demand may have risen well actually that may not be the case 183 00:19:00,640 --> 00:19:06,840 that would that would that would push prices up so I'm thinking to probably 184 00:19:06,840 --> 00:19:14,200 more supply side factors so things that would push supply up so either an 185 00:19:14,200 --> 00:19:19,800 increase in supply or better technology so it could just be that people in 186 00:19:19,800 --> 00:19:25,160 Oregon just have much much better cultivation technology than they do in 187 00:19:25,160 --> 00:19:32,720 Massachusetts and Maine it could be cheaper to produce cannabis in Oregon 188 00:19:32,720 --> 00:19:39,200 than it is to produce cannabis in Massachusetts and Maine right so people 189 00:19:39,200 --> 00:19:44,760 may be doing growing outdoor which may be harder to do in Massachusetts and 190 00:19:44,760 --> 00:19:51,320 Maine so that could be a factor as well I think we had a question coming from 191 00:19:51,320 --> 00:20:00,800 the audience yes and the question would be no um I raised my hand but I was just 192 00:20:00,800 --> 00:20:07,600 gonna add what Keegan just said it was very intellectual and you know it does 193 00:20:07,600 --> 00:20:14,640 make sense I know why it happened and Keegan said it but I'm I'm just 194 00:20:14,640 --> 00:20:21,840 wondering why Keegan is gonna say next I'm learning a lot that's why I raised my hand 195 00:20:21,840 --> 00:20:29,080 awesome and don't let me steal the show because ultimately it's a meetup right 196 00:20:29,080 --> 00:20:36,520 so you're right the idea is to have a conversation so don't let me steal the 197 00:20:36,520 --> 00:20:42,040 show I just I just put this together just to kind of have some guiding 198 00:20:42,040 --> 00:20:48,480 direction for the group so so if I ramble along too long and you've got a 199 00:20:48,480 --> 00:20:59,280 good comment then definitely chime in so so that was the main points I had for 200 00:20:59,280 --> 00:21:08,240 pricing so just it'll be interesting to observe how things go so I think we did 201 00:21:08,240 --> 00:21:15,840 forecasts of Oregon prices earlier in the year and we need to check those and 202 00:21:15,840 --> 00:21:23,040 I actually predicted that Oregon prices would increase through 2021 I called it 203 00:21:23,040 --> 00:21:30,440 you know inflation of cannabis prices but you're still seeing deflation of 204 00:21:30,440 --> 00:21:40,240 cannabis prices with prices going down in Oregon so real interesting and so only 205 00:21:40,240 --> 00:21:44,200 time will tell and so we'll just keep tracking these so we'll keep tracking 206 00:21:44,200 --> 00:21:49,920 Maine and Massachusetts and Oregon and add Washington to the mix and see if we 207 00:21:49,920 --> 00:22:01,120 can't add some more okay now time for the fun bits so we've got all this 208 00:22:01,120 --> 00:22:07,120 awesome sales data here and actually we can maybe predict prices so that could 209 00:22:07,120 --> 00:22:15,800 be a novel exercise but what I put together is predicting sales actually 210 00:22:15,800 --> 00:22:19,800 prices may be more interesting so we'll maybe try to save time to predict 211 00:22:19,800 --> 00:22:30,600 prices but basically it won't spend too much time on the code and so the data is 212 00:22:30,600 --> 00:22:37,480 more interesting than the code but long story short we're just going to iterate 213 00:22:37,480 --> 00:23:01,640 over the states we get the data for a specific state so I need to we'll get the 214 00:23:01,640 --> 00:23:14,800 data for a specific state Colorado data and then there's two ways to do this the 215 00:23:14,800 --> 00:23:21,880 first way I'm going to I'll just start with the month fixed effects and then 216 00:23:21,880 --> 00:23:28,720 show you the potential problem with those but basically I'm just going to 217 00:23:28,720 --> 00:23:42,720 add a dummy variable here for the month so that way we just have a zero or one 218 00:23:42,720 --> 00:23:56,440 depending on the month then I'm going to define the forecast horizon so basically 219 00:23:56,440 --> 00:24:04,960 we don't quite have dated through the end of the year so we only have data for 220 00:24:04,960 --> 00:24:11,280 Colorado through October so we'll actually have to forecast November 221 00:24:11,280 --> 00:24:24,480 December and then the 12 months of 2022 but we'll do that no no no problem and 222 00:24:24,480 --> 00:24:32,160 then we just have to forecast our month effects which we can do with a hundred 223 00:24:32,160 --> 00:24:39,440 percent accuracy because we always know what month it's going to be so this is a 224 00:24:39,440 --> 00:24:48,040 point if you're going to be adding fixed effects for forecasting you also have to 225 00:24:48,040 --> 00:24:55,640 forecast your your effects and so that can introduce you know additional error 226 00:24:55,640 --> 00:25:10,400 into your model using month is just a real easy one and then I'll show you why 227 00:25:10,400 --> 00:25:16,480 here in a second by why the downside but we're sort of going to be using the 228 00:25:16,480 --> 00:25:28,320 black box auto are my that PMD are my provides to the forecast and 229 00:25:31,200 --> 00:25:38,440 so the pros and the cons about using packages are the pros are right there 230 00:25:38,440 --> 00:25:50,280 are many built-in features and right there are many many contributors to get 231 00:25:50,280 --> 00:25:56,280 right to really iron out the code the downside is it's a bit of a black box and 232 00:25:56,280 --> 00:26:03,760 I am not a hundred percent certain it's using my fixed effects correctly and so 233 00:26:03,760 --> 00:26:12,480 long story short may end up sort of reading code to do this myself but for 234 00:26:12,480 --> 00:26:17,600 now we'll use the auto aremum package and I'll show you here in a second and 235 00:26:17,600 --> 00:26:26,600 I'm a little worried about it but long story short we're just using what's 236 00:26:26,600 --> 00:26:32,280 called the box Jenkins methodology which is you just take a series you look at 237 00:26:32,280 --> 00:26:41,880 its past behavior and you try to forecast future path of the variable and 238 00:26:41,880 --> 00:26:54,720 so this would be Colorado and for whatever reason it's not there's always 239 00:26:54,720 --> 00:27:05,880 a dip wherever whichever month I exclude so I wonder unless it's excluding 240 00:27:05,880 --> 00:27:34,360 informant but it shouldn't be so long story short is the right there's and 241 00:27:34,360 --> 00:27:46,000 then I'll show you what it's like to not include the month fix the text so for 242 00:27:46,000 --> 00:27:54,200 whatever reason now there's a large dip in April so there's something wrong with 243 00:27:54,200 --> 00:28:00,880 well maybe there's not necessarily wrong but I'm highly skeptical about what's 244 00:28:00,880 --> 00:28:05,440 going on with these fixed effects so I don't love it I do think it's important 245 00:28:05,440 --> 00:28:11,880 to control for things like the month effects but I'm not certain I'm doing it 246 00:28:11,880 --> 00:28:17,680 correctly so I just kind of wanted to show you that and then what would happen 247 00:28:17,680 --> 00:28:28,960 if you just you know excluded the month effects entirely well I probably don't 248 00:28:28,960 --> 00:28:39,680 need to run all of this well then we just use an autoregressive process and 249 00:28:39,680 --> 00:28:49,600 we don't have the dynamics that you would really expect out of a the truth 250 00:28:49,600 --> 00:28:58,320 like an actual series so this is with in just so that's sort of what's going on 251 00:28:58,320 --> 00:29:03,800 and just to kind of show you this in a better visualization with all the months 252 00:29:03,800 --> 00:29:09,600 now that you can kind of see what's going on actually let's do it first 253 00:29:09,600 --> 00:29:16,200 without the month fixed effects so here just going to iterate over all of the 254 00:29:16,200 --> 00:29:27,280 states and do this forecast and then down here I've plotted it in a decent 255 00:29:27,280 --> 00:29:31,880 plot so sorry that I'm rushing through the code but I think the data and 256 00:29:31,880 --> 00:29:38,920 visualizations are going to be more interesting so here you can kind of see 257 00:29:38,920 --> 00:29:54,800 the forecast with the in here I can make this larger so here you can see just if 258 00:29:54,800 --> 00:30:06,840 we just use the autoregressive forecast without any month effects then the most 259 00:30:06,840 --> 00:30:16,680 recent observations and direction really influences our prediction so for example 260 00:30:16,680 --> 00:30:24,440 in Colorado because the last few observations were decreasing we we 261 00:30:24,440 --> 00:30:34,200 predict a decreasing trend similarly with Washington peculiar peculiar the 262 00:30:34,200 --> 00:30:43,520 we predicted outwards trend in Illinois Massachusetts super tough to predict 263 00:30:43,520 --> 00:30:57,280 because it's just had this you know volatile number of sales and so we know 264 00:30:57,280 --> 00:31:05,480 our forecast is really naive and it's kind of based on this last dip but you 265 00:31:05,480 --> 00:31:13,120 know we may actually see Massachusetts you know return to a to a higher 266 00:31:13,120 --> 00:31:19,520 level and then you know we were not predicting too much change in Maryland 267 00:31:19,520 --> 00:31:25,080 maybe a slight positive increase and then you know we were predicting a 268 00:31:25,080 --> 00:31:36,160 drop-off in Maine which also I'm not quite certain about so those would just 269 00:31:36,160 --> 00:31:44,840 be the autoregressive trends that we would forecast and then if we add in the 270 00:31:44,840 --> 00:31:56,000 month effects which I say I'm not very certain about they you'll see they we 271 00:31:56,000 --> 00:32:04,600 seem to give more plausible forecasts except for the month excluded so you 272 00:32:04,600 --> 00:32:11,840 know so something's some fix is going to need to happen to better account for 273 00:32:11,840 --> 00:32:18,240 that but that's why remember our you know ten commandments of forecasting you 274 00:32:18,240 --> 00:32:29,040 know it's in the forecasting models are gonna evolve over time so so now that 275 00:32:29,040 --> 00:32:37,680 you've been warned about but this model is far far from perfect and because 276 00:32:37,680 --> 00:32:44,360 something is going on here here or you know essentially forecast for the various 277 00:32:44,360 --> 00:32:54,760 states and here if you include the month effects besides from this unrealistic 278 00:32:54,760 --> 00:33:03,440 drop in January which I need to fix I think it's something to do with how the 279 00:33:03,440 --> 00:33:14,560 model the models predicting great question yeah Keegan um um from my time 280 00:33:14,560 --> 00:33:21,600 as a data scientist this looks a lot like artificial error and stuff what 281 00:33:21,600 --> 00:33:26,760 you're saying basically stuff with the model and how you put in the month 282 00:33:26,760 --> 00:33:37,680 excluded I'm not sure what's happening my best guess is that sales data keeps 283 00:33:37,680 --> 00:33:45,520 some sort of running some in its forecast and the plus or minus on each 284 00:33:45,520 --> 00:33:55,600 month each Delta T state is what changes month to month and when you exclude that 285 00:33:55,600 --> 00:34:04,560 month you also exclude that running total so that month exclusion principle 286 00:34:04,560 --> 00:34:14,800 maybe excluding that running total as well I'm not sure but I'm gonna make a 287 00:34:14,800 --> 00:34:25,080 hard guess that it has something to do with how it handles that month excluded 288 00:34:25,080 --> 00:34:36,040 in Europe 100% correct and so unfortunately actually I can actually 289 00:34:36,040 --> 00:34:42,040 tinker with this and make sure that this is correct so watch for an e-message 290 00:34:42,040 --> 00:34:49,880 from me later today if I fix this but something's going on I actually wrote 291 00:34:49,880 --> 00:35:01,280 this function myself so I can double check to see if the breakdown was in 292 00:35:01,280 --> 00:35:07,360 there but I've got a bad feeling I'm gonna have to write the Norema model 293 00:35:07,360 --> 00:35:19,240 with exogenous effects which is not the end of the world but you know that's why 294 00:35:19,240 --> 00:35:25,840 I first you know reach for some packages but a lot of times with statistics it 295 00:35:25,840 --> 00:35:33,600 doesn't hurt to do it yourself because statistics are high-stake and they're 296 00:35:33,600 --> 00:35:43,200 hard to get right so that was what I did a lot when I was in school is it was 297 00:35:43,200 --> 00:35:48,360 basically just writing just kind of double checking statistics that you know 298 00:35:48,360 --> 00:35:53,360 these statistical packages produce you know making sure the variances are 299 00:35:53,360 --> 00:36:00,520 correct and this and that in like I said you know night probably 95% of the time 300 00:36:00,520 --> 00:36:07,600 you're going to be golden but but something's going on here and I'm not 301 00:36:07,600 --> 00:36:17,680 super confident about these forecasts so so it's unfortunate but as you've learned 302 00:36:17,680 --> 00:36:25,000 I don't you the cannabis data science group we never really let these hurdles 303 00:36:25,000 --> 00:36:30,640 slow us down or at least not too much and so we're going to acknowledge that 304 00:36:30,640 --> 00:36:38,240 our forecast in January is messed up and then that may even bias our future 305 00:36:38,240 --> 00:36:46,440 result or future forecasts as well so our forecasts are going to be biased but 306 00:36:46,440 --> 00:36:54,040 any number well not necessarily but in this case I feel that any number is 307 00:36:54,040 --> 00:37:01,000 better than no number so long story short it would just be nice to have a 308 00:37:01,000 --> 00:37:11,320 number for what 2022 sales may be even if you know it may be wildly off and we 309 00:37:11,320 --> 00:37:20,240 revise our forecast in the future so so long story short this model needs to get 310 00:37:20,240 --> 00:37:30,200 fixed so keep that in mind with with the numbers about to show you but I think 311 00:37:30,200 --> 00:37:39,120 we're on the right track if we can correct January then we get plausible 312 00:37:39,120 --> 00:37:50,200 fluctuation and we even we're even able to to get Massachusetts to the 313 00:37:50,200 --> 00:37:59,160 to have its its unique cyclical trend so to the month effects I think are 314 00:37:59,160 --> 00:38:11,200 important but just need to get fixed so with that said we can the last bit was 315 00:38:11,200 --> 00:38:19,000 basically to analyze our forecasts so I thought the simplest way to do it would 316 00:38:19,000 --> 00:38:30,840 just be to count the amount of sales that will happen in 2022 so just to run 317 00:38:30,840 --> 00:38:35,640 that bit of code for you sorry I'm rushing through the code but the nice 318 00:38:35,640 --> 00:38:48,720 thing is it'll be on github to double check and you know to read through so 319 00:38:48,720 --> 00:39:02,480 even with our are not very believable forecasted dip in January you know 320 00:39:02,480 --> 00:39:08,160 we're still predicting and so basically what I'm going to say is I think all of 321 00:39:08,160 --> 00:39:14,800 these estimates are biased downwards because I don't think we're capturing 322 00:39:14,800 --> 00:39:22,000 January sales correctly so so I think these are are going to be sort of my 323 00:39:22,000 --> 00:39:31,800 like lower limit forecasts for these states so you've heard it here December 324 00:39:31,800 --> 00:39:38,520 22nd 2021 Keegan's key and the cannabis data science meetup group made these 325 00:39:38,520 --> 00:39:48,960 predictions and so I'm predicting that Colorado will have at least 1.6 billion 326 00:39:48,960 --> 00:39:59,400 in sales in 2022 this may be lower than other people's forecasts I've heard some 327 00:39:59,400 --> 00:40:09,520 forecasts for Colorado as much as 2 billion per in 2022 so so my forecast is 328 00:40:09,520 --> 00:40:19,320 a little lower than that 1.6 but it's biased downwards Illinois predicting 329 00:40:19,320 --> 00:40:33,080 will be just shy of 1 billion in sales in 2022 so 934 million that's a in I'll 330 00:40:33,080 --> 00:40:38,160 try to put that into perspective here in a minute because these are really large 331 00:40:38,160 --> 00:40:48,680 figures and it's hard to conceptualize the this amount of money so I'm going to 332 00:40:48,680 --> 00:40:59,720 put that in perspective here so we'll do that in a second Massachusetts 680 333 00:40:59,720 --> 00:41:10,120 million Maryland actually not that far behind and so that's why we may want to 334 00:41:10,120 --> 00:41:19,520 I wouldn't be surprised if Massachusetts comes in a lot above 680 million in 335 00:41:19,520 --> 00:41:33,760 2022 but at least we have an estimate main around 74 million Oregon around 1 336 00:41:33,760 --> 00:41:43,040 billion in 2022 this is biased downwards I think and then Washington around 1.2 337 00:41:43,040 --> 00:41:52,320 billion and so we have the the populations of the various states so you 338 00:41:52,320 --> 00:42:06,640 know you can start to see well well we could actually potentially look at that 339 00:42:06,640 --> 00:42:13,360 because basically sales per person would be sort of GDP per person and that would 340 00:42:13,360 --> 00:42:25,800 be about how much better off people in these states are so actually so a couple 341 00:42:25,800 --> 00:42:40,560 of statistics said these will just be kind of for fine so so one thing that 342 00:42:40,560 --> 00:42:51,240 metric that I'd like to to use is number of schools and so in my hometown a few 343 00:42:51,240 --> 00:42:57,120 years back they built a new school for I want to say two million dollars that 344 00:42:57,120 --> 00:43:05,120 sounds low but I swear that's what it was but you know let's say you know a 345 00:43:05,120 --> 00:43:14,800 new school you know costs you know five million dollars so now we could you 346 00:43:14,800 --> 00:43:23,840 know start to you know measure you know sales as in terms of you know the number 347 00:43:23,840 --> 00:43:37,680 of new schools so but of course you know you can't a tax people a hundred percent 348 00:43:37,680 --> 00:43:44,640 so you know you could just say oh you know what happens if you know you did 349 00:43:44,640 --> 00:43:54,040 you know a tax of just say like seven percent or something so this long story 350 00:43:54,040 --> 00:43:59,040 short is that I'm sort of doing this as a lesson for some of the states that may 351 00:43:59,040 --> 00:44:05,320 not have permitted adult use yet we'll just kind of do a thought exercise here 352 00:44:05,320 --> 00:44:12,920 and just say okay if you taxed all of this revenue at seven percent how many 353 00:44:12,920 --> 00:44:23,320 you know new schools could you could you build in 2022 so let's see if we can't 354 00:44:23,320 --> 00:44:34,240 do this I'm just kind of spitballing this here just for fun so let's see if 355 00:44:34,240 --> 00:44:43,880 we can't do that so you're going to get that amount of sales and they'll tax it 356 00:44:43,880 --> 00:44:57,880 at seven percent then you'll divide that by the cost of a new school so so so 357 00:44:57,880 --> 00:45:16,160 that state you know could build with many you know new schools but let's do 358 00:45:16,160 --> 00:45:22,400 this first and then we'll we'll look at sort of the GDP per person so let's 359 00:45:22,400 --> 00:45:31,720 just try this like I said I'm just sort of just sort of spitballing this here 360 00:45:31,720 --> 00:45:38,320 but you know so we're saying you know a new school cost five million which I 361 00:45:38,320 --> 00:45:43,160 don't know if that's accurate or not but let's say that's the price of a really 362 00:45:43,160 --> 00:45:50,480 nice school well you know Colorado you know if they just and of course all of 363 00:45:50,480 --> 00:45:57,360 these states are already taxed but you know like I said this is sort of a thought 364 00:45:57,360 --> 00:46:04,120 exercise for for some new states and there may be thinking about what to set 365 00:46:04,120 --> 00:46:08,680 their tax rates at and what they may want to spend their money on and how far 366 00:46:08,680 --> 00:46:17,480 it'll go well let's say you know your state rate and we have the population 367 00:46:17,480 --> 00:46:27,720 here so population in Colorado in the latest check was five point eight million 368 00:46:27,720 --> 00:46:39,280 people so you know you could build you know 23 new schools so I think that's 369 00:46:39,280 --> 00:46:45,480 just sort of a you know a way that you could start to you kind of quantify this 370 00:46:45,480 --> 00:46:52,320 and I think you know kind of argue you know so for example in you know like 371 00:46:52,320 --> 00:46:57,360 Maryland you could say hey you know we only have medicinal you know and we can 372 00:46:57,360 --> 00:47:03,680 only make eight new schools you know maybe if we allowed adult use you know 373 00:47:03,680 --> 00:47:09,360 eventually we could be building you know 23 new schools a year like Colorado can 374 00:47:09,360 --> 00:47:17,480 do so so I just thought that was an interesting metric but just really 375 00:47:17,480 --> 00:47:26,840 arbitrary just kind of pulling numbers out of my hat so nothing special about 376 00:47:26,840 --> 00:47:36,080 that just trying to put some of these numbers in perspective and then let's 377 00:47:36,080 --> 00:47:44,840 see if there's any elegant way to get the sorry to do this coding in front of 378 00:47:44,840 --> 00:47:56,720 you but let's see if there's any way to get the latest population so oh and I 379 00:47:56,720 --> 00:48:01,200 forgot that we were going to try to predict prices so that may have to wait 380 00:48:01,200 --> 00:48:09,480 till next week but you can at least try to quantify how much better off people 381 00:48:09,480 --> 00:48:20,640 in these states are going to be okay so we're gonna go ahead and get that series 382 00:48:20,640 --> 00:48:35,040 and basically we want the last population know if this is going to be 383 00:48:35,040 --> 00:48:44,480 okay so that's how we can get the latest population so we can get you know 384 00:48:44,480 --> 00:48:56,240 essentially GDP per capita in these states for cannabis GDP so you know the 385 00:48:56,240 --> 00:49:02,320 government's always measuring GDP well we can kind of you know show how much 386 00:49:02,320 --> 00:49:12,480 better off you know people are from from candidates and so found this just do it 387 00:49:12,480 --> 00:49:21,000 like that and no promises this is going to work 388 00:49:24,680 --> 00:49:34,240 and GDP per capita from cannabis and no promises this is going to work first 389 00:49:34,240 --> 00:49:47,560 try but we can always try okay so looks like me we may have messed up our 390 00:49:47,560 --> 00:49:55,560 calculation but I always think these numbers look low but you have to 391 00:49:55,560 --> 00:50:03,120 remember this is this is everybody in the state you know everyone from newborn 392 00:50:03,120 --> 00:50:12,200 to retired person so like in Colorado you know every like everybody's going to 393 00:50:12,200 --> 00:50:21,600 be about you know almost $300 better off because because cannabis is legalized 394 00:50:21,600 --> 00:50:29,000 and those would be people that aren't even consuming cannabis so this would 395 00:50:29,000 --> 00:50:34,080 just this is just more money in the economy and you know we haven't even 396 00:50:34,080 --> 00:50:40,720 really taken into consideration there could be a multiplier effect so people 397 00:50:40,720 --> 00:50:45,400 that run cannabis businesses they employ other businesses who in turn employ 398 00:50:45,400 --> 00:50:53,360 other businesses so there's sort of a multiplier effect when what you have the 399 00:50:53,360 --> 00:50:59,400 you know these number of sales and remember you know these are sales that 400 00:50:59,400 --> 00:51:05,280 may have otherwise been taking place you know below ground you know not getting 401 00:51:05,280 --> 00:51:12,640 taxed with people who are willing to operate below ground getting this amount 402 00:51:12,640 --> 00:51:17,080 of money and I'm not saying that you know everybody's necessarily a bad 403 00:51:17,080 --> 00:51:26,360 character but I mean this is a lot of money to be you know flowing untracked 404 00:51:26,360 --> 00:51:33,320 in the economy to potentially some shady characters right like we can't pretend 405 00:51:33,320 --> 00:51:38,960 that they're going to be no shady characters in the black market of 406 00:51:38,960 --> 00:51:45,720 cannabis and so that's sort of what my sort of takeaway from today is and today 407 00:51:45,720 --> 00:51:51,560 it's always more of a lesson to other states and so although we looked at 408 00:51:51,560 --> 00:51:59,520 these states I think this is here let's let's put it all together I think this 409 00:51:59,520 --> 00:52:09,560 is more of a thought exercise for some of the states that may be on the fence 410 00:52:09,560 --> 00:52:19,400 about cannabis because I mean the alternative is you either have a black 411 00:52:19,400 --> 00:52:25,080 market in your state with potentially millions of dollars getting into the 412 00:52:25,080 --> 00:52:33,360 hands of people that you don't have the least idea about and the alternative 413 00:52:33,360 --> 00:52:39,920 your opportunity cost is schools right your opportunity cost of keeping 414 00:52:39,920 --> 00:52:47,960 cannabis illegal is potentially building you know up to you know 23 or so schools 415 00:52:47,960 --> 00:52:59,800 and remember these are are low estimates so we're gonna revise our forecasts here 416 00:52:59,800 --> 00:53:08,200 to really try to get as close to the mark as possible but I mean I can't 417 00:53:08,200 --> 00:53:14,760 think this is anything but a good thing for these states I mean Illinois can 418 00:53:14,760 --> 00:53:24,200 potentially build 13 you know top-notch new schools next year then the 419 00:53:24,200 --> 00:53:33,920 alternative would have been one billion dollars you know flowing into you know 420 00:53:33,920 --> 00:53:43,840 who knows whose hands so I think that's sort of my main point for today so I 421 00:53:43,840 --> 00:53:49,720 guess I just wanted to share with you sort of any of you are you know speaking 422 00:53:49,720 --> 00:53:56,120 to you know your state legislatures or just you know your community because 423 00:53:56,120 --> 00:54:00,720 kind of things like this spread so like you know kind of talk about this with 424 00:54:00,720 --> 00:54:07,520 with your friends and family and explain to them like hey like not only are they 425 00:54:07,520 --> 00:54:14,320 making a large amount of money but you can all put it into concrete terms where 426 00:54:14,320 --> 00:54:21,600 you know these states could be building dozens and dozens of new schools each 427 00:54:21,600 --> 00:54:30,560 year that's just 2022 just 2022 I mean you could just in then and these are 428 00:54:30,560 --> 00:54:40,480 growing series so I forgot to mention that it looks like Colorado where 429 00:54:40,480 --> 00:54:46,040 Colorado could have peaked but for the most part these are growing series and 430 00:54:46,040 --> 00:55:03,400 so you know if I was a state so I was the state you know so here I am in North 431 00:55:03,400 --> 00:55:12,280 Carolina population 10.6 million people you know I'd be looking at some of these 432 00:55:12,280 --> 00:55:19,000 other states and be saying like hey like you know Colorado has a population of 433 00:55:19,000 --> 00:55:29,360 5.8 million and they can build 23 new schools like you know how many schools 434 00:55:29,360 --> 00:55:36,280 could we potentially build or Texas Texas has a population of 30 million 435 00:55:36,280 --> 00:55:43,720 that's or almost 30 million right that's well I guess Colorado is closer to 6 436 00:55:43,720 --> 00:55:50,640 million so let's say Texas is about five times the size of Colorado well Texas if 437 00:55:50,640 --> 00:55:58,120 you legalize adult use cannabis you know you're not going to get there overnight 438 00:55:58,120 --> 00:56:05,360 but you know you could potentially be building five times the number of 439 00:56:05,360 --> 00:56:12,120 schools in Colorado that's a hundred schools per year so if you're you know a 440 00:56:12,120 --> 00:56:18,960 resident of Texas I mean these are sort of the things that you can try to use to 441 00:56:18,960 --> 00:56:26,000 persuade people who are on the fence and say like hey you know it's either this 442 00:56:26,000 --> 00:56:32,000 is your opportunity cost it's you can either let you know millions if not 443 00:56:32,000 --> 00:56:39,800 billions of dollars flow into the hands of who knows who or you could build you 444 00:56:39,800 --> 00:56:46,240 know a hundred new schools per year in Texas you know so these are sort of a 445 00:56:46,240 --> 00:56:53,120 sort of persuasion tools that are arming you with and so you know you can look to 446 00:56:53,120 --> 00:57:07,640 states that are doing well so and then and see you know how their policies may 447 00:57:07,640 --> 00:57:13,640 work or may not work and so so I think that's sort of my main lesson for today 448 00:57:13,640 --> 00:57:17,960 but I kind of want to end it there and see if anyone has any thoughts but but 449 00:57:17,960 --> 00:57:22,560 that was sort of what we've been building up to this year is when we had 450 00:57:22,560 --> 00:57:31,400 to get the data then we had to analyze it and make our forecasts and now we can 451 00:57:31,400 --> 00:57:37,320 sort of discuss and analyze our forecasts and so the final statistic I'll 452 00:57:37,320 --> 00:57:47,160 leave you with here today is keep in mind we've we've excluded a couple states 453 00:57:47,160 --> 00:57:54,320 here from our analysis so we still need to look at California the most populous 454 00:57:54,320 --> 00:58:00,520 state almost 40 million people and we don't have as a single good data point 455 00:58:00,520 --> 00:58:07,320 yet and so we're working on it I just just haven't looked at California yet 456 00:58:07,320 --> 00:58:14,720 just by happenstance but we still need to you know get Arizona data in the 457 00:58:14,720 --> 00:58:18,920 process of seeing if we can't get Michigan data which is almost 10 million 458 00:58:18,920 --> 00:58:26,880 people that's a that's comparable to North Carolina so and actually still need 459 00:58:26,880 --> 00:58:32,400 to get Montana's data 1 million people there can't can't can't count them out 460 00:58:32,400 --> 00:58:38,760 and so so there's still a few more states to add but even with this all the 461 00:58:38,760 --> 00:58:47,560 states that we have calculated we've calculated that there'll be at least 6 462 00:58:47,560 --> 00:58:56,920 billion in cannabis sales in 2022 just in you know these the just from these 463 00:58:56,920 --> 00:59:04,760 handful of states not there will be we predict there will be so so 6 billion in 464 00:59:04,760 --> 00:59:12,480 sales well you know uncle Sam may want to you know start start looking at this 465 00:59:12,480 --> 00:59:19,280 too right because six that's six billion in well actually I guess these people 466 00:59:19,280 --> 00:59:23,960 already are paying taxes so actually they may already be paying tax to uncle 467 00:59:23,960 --> 00:59:30,000 Sam but long story short is you know maybe even the federal government may 468 00:59:30,000 --> 00:59:38,280 want to start thinking about legalization because once again you know 469 00:59:38,280 --> 00:59:46,280 they may not spend the money on schools per se but you know there's a lot of I 470 00:59:46,280 --> 00:59:53,120 think there's a lot of social good that can be that can be had and so you're now 471 00:59:53,120 --> 00:59:59,760 armed with the data some economic theory and some statistics and some know-how 472 00:59:59,760 --> 01:00:05,040 about the cannabis industry so I hope you all go and you know spread the word 473 01:00:05,040 --> 01:00:16,680 and you know keep advancing cannabis science on your own as well so we're 474 01:00:16,680 --> 01:00:21,640 gonna keep this up so as I said you know here in 2022 we'll be checking all of 475 01:00:21,640 --> 01:00:27,880 these forecasts right because if we're wildly off we want to to admit that and 476 01:00:27,880 --> 01:00:33,400 let's show show everybody how wrong we were so so that's something to look 477 01:00:33,400 --> 01:00:38,760 forward to and then the other thing is in 2022 we've been so focused on sales 478 01:00:38,760 --> 01:00:43,080 this year we're gonna be digging a lot more into the production side of things 479 01:00:43,080 --> 01:00:48,400 so that way we can start looking at cannabinoids and turpines which were 480 01:00:48,400 --> 01:00:54,040 some of the things we were set out to look at and then we can start looking at 481 01:00:54,040 --> 01:00:58,880 some more of the cultivation metrics maybe even start to analyze 482 01:00:58,880 --> 01:01:05,280 manufacturing because manufacturing something I know the least about in the 483 01:01:05,280 --> 01:01:14,760 cannabis industry so it can be a fun time to learn together so I think that's 484 01:01:14,760 --> 01:01:18,920 where I'm gonna leave it for today don't want to take up too much of your time I 485 01:01:18,920 --> 01:01:24,520 know everyone's time is precious so any thoughts or comments before we conclude 486 01:01:24,520 --> 01:01:26,960 for today 487 01:01:30,680 --> 01:01:37,880 so I just think it's a good job it was it was really informative um you did a 488 01:01:37,880 --> 01:01:45,960 good job of just explaining these are for fun statistics at the end but it was 489 01:01:45,960 --> 01:01:53,360 very informative and educational and I'm excited for the cannabinoids because I've 490 01:01:53,360 --> 01:02:03,680 been messing around with data science with cannabinoids for a little bit and I 491 01:02:03,680 --> 01:02:11,760 would love to hear insights on that as well definitely Graham and in fact if 492 01:02:11,760 --> 01:02:17,800 you want to share any of your work you're more than welcome to so we've only 493 01:02:17,800 --> 01:02:23,200 been able to look at turpene and maybe cannabinoid data from Connecticut so 494 01:02:23,200 --> 01:02:28,440 there's a good data set there that we've been working with so let's 495 01:02:28,440 --> 01:02:35,880 definitely go back and forth and start preparing some good scientific analysis 496 01:02:35,880 --> 01:02:40,720 for the coming weeks because we've gotten so tied in with economics let's 497 01:02:40,720 --> 01:02:46,400 get to some some interesting scientific questions so I think there is there's so 498 01:02:46,400 --> 01:02:52,800 much to cover so so definitely stay tuned for some exciting work there 499 01:02:52,800 --> 01:02:58,560 awesome awesome well I'm going to go ahead and conclude it for today just so 500 01:02:58,560 --> 01:03:03,560 you know Saturday morning statistics is on Christmas so you don't necessarily 501 01:03:03,560 --> 01:03:08,380 have to attend but I was thinking if you want to sign up and register so it's 502 01:03:08,380 --> 01:03:13,600 just one dollar you can still get all of the material so you don't have to 503 01:03:13,600 --> 01:03:20,920 actually attend you can just you know enjoy festivities or what have you and I 504 01:03:20,920 --> 01:03:25,560 can send you the recording and the material afterwards so just thought I 505 01:03:25,560 --> 01:03:28,840 would throw that out there so if any of you do want to sign up for Saturday 506 01:03:28,840 --> 01:03:34,240 morning statistics I always try to make it worth your while so for one dollar I 507 01:03:34,240 --> 01:03:41,720 think you'll get I think hopefully you'll get more than that in value so 508 01:03:41,720 --> 01:03:46,560 just thought I would promote that real quick so until next Wednesday or 509 01:03:46,560 --> 01:03:51,200 Saturday if you want to tune in I hope you all have a productive week keep your 510 01:03:51,200 --> 01:04:05,400 move to the grindstone and have fun