1 00:00:00,000 --> 00:00:10,160 Welcome to cannabis data science. You're in the right place. You're in for one of the 2 00:00:10,160 --> 00:00:16,120 biggest treats I think the group has to offer yet to date. So the cannabis data science 3 00:00:16,120 --> 00:00:22,780 team has been working on many different tasks and you know my background from the laboratory. 4 00:00:22,780 --> 00:00:27,720 So we've got a little bit of know how and we've got some awesome programmers such as 5 00:00:27,720 --> 00:00:38,600 Candice who's got a long career in IT. So we've got a fantastic team here and essentially 6 00:00:38,600 --> 00:00:44,000 we'll go ahead and let the cat out of the bag. The task that's been on everybody's mind 7 00:00:44,000 --> 00:00:48,680 for a long time now and in fact this was one of the reasons that Canelytics was started 8 00:00:48,680 --> 00:00:56,480 was wanted to get lab results in consumers hands. So in the cannabis industry we've got 9 00:00:56,480 --> 00:01:02,640 mandated quality control testing. So that's good for the consumers right. So products 10 00:01:02,640 --> 00:01:10,160 get screened for contaminants. They get tested for cannabinoids so you know what you're consuming. 11 00:01:10,160 --> 00:01:15,960 But the data I mean how often do you actually get your hands on the data. And so we're some 12 00:01:15,960 --> 00:01:22,360 of the most avid pursuers of this data and it's still hard to come by. So your average 13 00:01:22,360 --> 00:01:28,440 consumer may have a really really tough time with this. So that's one of the reasons Canelytics 14 00:01:28,440 --> 00:01:33,240 was started. So we were trying to work with labs, producers, processors, retailers trying 15 00:01:33,240 --> 00:01:38,560 to figure out how can we get this data to the consumers. And then finally there's been 16 00:01:38,560 --> 00:01:47,300 some awesome fine folks. And so I want to really commemorate John. So John and the CESC. 17 00:01:47,300 --> 00:01:57,800 So check out and want to say it's dcesc.org has been moving the ball forward for the consumers 18 00:01:57,800 --> 00:02:02,160 right. So they're trying to get the consumers into the picture right. At the end of the 19 00:02:02,160 --> 00:02:06,400 day they're the ones driving the industry. So we want to make sure that people can make 20 00:02:06,400 --> 00:02:14,400 healthy, wise, educated decisions. And then you know they know what they're in store for 21 00:02:14,400 --> 00:02:19,600 and you know we need to get that data and then get that into consumers hands. So really 22 00:02:19,600 --> 00:02:25,200 want to say that work that John's doing is noble and that the task that John sent me 23 00:02:25,200 --> 00:02:35,600 on was the labs and the producers. One can almost say you know they've taken the ball 24 00:02:35,600 --> 00:02:41,160 as far as they can and so there's just somebody who needs to go and take it you know the last 25 00:02:41,160 --> 00:02:48,120 10 yards. And that's sort of the Cantlydix philosophy is we don't like impose upon anyone 26 00:02:48,120 --> 00:02:52,880 or we try not to. We try to meet you where you are and take you one step further. And 27 00:02:52,880 --> 00:03:05,600 so where people are is they've got COAs. And in fact I've got this is the first day I've 28 00:03:05,600 --> 00:03:14,480 actually had props. So instead of a presentation I brought props today. So at the cannabis 29 00:03:14,480 --> 00:03:23,240 dispensary you get these items. And I was saying the last week that the QR codes were 30 00:03:23,240 --> 00:03:28,560 20th century technology and I may want to walk that back. Technically they are. They 31 00:03:28,560 --> 00:03:35,720 were I think invented in like 1994 1995 or so. But that's late 20th century technology 32 00:03:35,720 --> 00:03:40,520 and I think they're just getting put to good use today. So essentially you know with these 33 00:03:40,520 --> 00:03:47,160 QR codes at the dispensary you can get the COA and we can show you that today. But then 34 00:03:47,160 --> 00:03:53,940 it's like well now we actually need that data. And so that is the task that we've finally 35 00:03:53,940 --> 00:04:02,640 solved or at least begun to solve as you'll see there's almost infinite room for improvement 36 00:04:02,640 --> 00:04:13,080 as always. But I think we've done a lot here. So before we jump into it any thoughts or 37 00:04:13,080 --> 00:04:22,200 I guess I'm still kind of building up suspense here. Well without further ado let's just 38 00:04:22,200 --> 00:04:32,920 go ahead and start the presentation because you're both going to be in for quite the treat 39 00:04:32,920 --> 00:04:45,760 here. We're going to need some tools for the job here. So basically where do we even begin 40 00:04:45,760 --> 00:04:58,400 with this? Well let's look at one of these COAs. So if you were to scan one of these 41 00:04:58,400 --> 00:05:05,920 QR codes say you got a QR code reader on your phone which I do but just for the sake of 42 00:05:05,920 --> 00:05:13,040 the presentation let's just pretend okay I just scanned this QR code on my phone and 43 00:05:13,040 --> 00:05:22,560 then this is going to shoot you over to a URL. And in this case we've got a Confident 44 00:05:22,560 --> 00:05:30,860 Cannabis URL and Confident Cannabis is a common laboratory information management systems 45 00:05:30,860 --> 00:05:39,560 or LIMS provider. And so a lot of laboratories will use Confident Cannabis and they'll have 46 00:05:39,560 --> 00:05:47,880 their online portal here where you can see a picture, you can see the cannabinoids, you 47 00:05:47,880 --> 00:05:59,480 can see the screenings. So no heavy metals in this flower. Awesome. And even better no 48 00:05:59,480 --> 00:06:06,760 pesticides detected. And so this is really what you want to see. So especially I would 49 00:06:06,760 --> 00:06:12,320 say if you're consuming concentrates you probably want to check your solvent cell to 50 00:06:12,320 --> 00:06:18,200 make sure you've got a nice clean concentrate. So long story short you can get the results 51 00:06:18,200 --> 00:06:30,800 but we want to actually get these physical data points. And so this is quality control 52 00:06:30,800 --> 00:06:38,760 tested mandated by the state, the laboratory provided that data and now the data is here 53 00:06:38,760 --> 00:06:46,960 publicly available on the web. So these are now facts. So this is now essentially a fact 54 00:06:46,960 --> 00:06:55,800 about this classic JAK sample. So readily collectible at this point. And in fact I think 55 00:06:55,800 --> 00:07:03,680 as a consumer you should collect your cannabis data because I heard a saying once that if 56 00:07:03,680 --> 00:07:10,760 you're not measuring it you're not managing it. So if you want to manage your cannabis 57 00:07:10,760 --> 00:07:19,160 consumption then you'll need to measure it. And these days with the cannabinoid concentrations 58 00:07:19,160 --> 00:07:25,760 getting quite potent you really want to be keeping track of the milligrams because just 59 00:07:25,760 --> 00:07:36,400 one gram of cannabis is not the same across the board. Cool. So that's the work that's 60 00:07:36,400 --> 00:07:45,280 going to be had and just to kind of knock out two birds with one stone let's say you 61 00:07:45,280 --> 00:07:58,880 also have this PDF. So sometimes the laboratory may only have this PDF. So we'll also save 62 00:07:58,880 --> 00:08:07,200 a copy of this. That's one of the ones I sent you I think right. I think this is actually 63 00:08:07,200 --> 00:08:14,360 one from a recent purchase of mine at a cannabis dispensary. It's at Cali Green. So did that 64 00:08:14,360 --> 00:08:22,840 mean meanwhile you were down here. Exactly. So basically I've been saving some of these 65 00:08:22,840 --> 00:08:31,600 packaging packages because as I was saying like and this is why I think this is novel 66 00:08:31,600 --> 00:08:40,680 right. You know I'm a cannabis consumer and I'm interested in keeping track of this data. 67 00:08:40,680 --> 00:08:49,760 And if you know myself who I think is you know I try to be a a practitioner data scientist 68 00:08:49,760 --> 00:08:57,880 you know that's up for you to debate if I am or not but I try to do data science. And 69 00:08:57,880 --> 00:09:06,600 so if I can't even get a hold of my own lab results you know then how could you know how 70 00:09:06,600 --> 00:09:12,480 could like any you know it's going to be it's going to be tough for consumers to get their 71 00:09:12,480 --> 00:09:19,920 their lab results. So long story short is let's solve this problem. Let's solve it today. 72 00:09:19,920 --> 00:09:23,920 And then we'll basically do a proof of concept and then I'll show you how this can be iterated 73 00:09:23,920 --> 00:09:35,880 upon to you know further get refined and incorporated into many many applications. So just so I've 74 00:09:35,880 --> 00:09:45,560 downloaded some C.A. ways. So here's this same C.A. way the Cali green. It was to see 75 00:09:45,560 --> 00:09:58,960 here you see it was tested at the Cali green laboratory by confident cannabis. And so basically 76 00:09:58,960 --> 00:10:08,480 to to just show you how the magic is going to happen instead of just pretending things 77 00:10:08,480 --> 00:10:16,480 are magic. Right. So we've got PDF plumber. And so my first go of this was OK. Let's find 78 00:10:16,480 --> 00:10:24,600 all the tables on here and just start parsing them. And it's a mess. And I think it can't 79 00:10:24,600 --> 00:10:33,840 be done and probably should be done. But one thing I learned work Tory was there was always 80 00:10:33,840 --> 00:10:42,480 this stress that OK you need to design things to be compatible with being offline. But any 81 00:10:42,480 --> 00:10:51,280 time we were offline the number one goal was getting back online. So it's basically like 82 00:10:51,280 --> 00:10:58,360 if you're not online you know maybe not. Don't be worrying about parsing C.A. ways like you 83 00:10:58,360 --> 00:11:02,360 know you should first worry about getting online because you need to check your emails 84 00:11:02,360 --> 00:11:07,040 and other things. And then once you're online then you can you know parse C.A. ways to your 85 00:11:07,040 --> 00:11:14,460 heart's content. So we can worry about offline in the future. But instead of parsing all 86 00:11:14,460 --> 00:11:19,560 of this let's just be smart about this. Right. So we're going going and going. And so here 87 00:11:19,560 --> 00:11:25,280 I'm parsing I'm like I'm getting the images that that that that that. And then I realize 88 00:11:25,280 --> 00:11:37,720 oh wait a second. This PDF has a QR code on it. So what can we do with that is I was thinking 89 00:11:37,720 --> 00:11:49,360 no way would it be possible to essentially read the QR code programmatically and typically 90 00:11:49,360 --> 00:11:57,720 with Python. If it's on a computer where there's a will there's a way. That's what I've that's 91 00:11:57,720 --> 00:12:04,160 why I love Python. That's kind of why I fell in love with it is I started realizing that 92 00:12:04,160 --> 00:12:15,440 anything on the computer you could essentially automate with Python. And I just want to give 93 00:12:15,440 --> 00:12:22,360 a quick shout out to the Selenium authors because this is a package that I've kind of 94 00:12:22,360 --> 00:12:28,720 shied away from because it's got a bad reputation in the Python community. We're not a bad reputation 95 00:12:28,720 --> 00:12:36,200 but it's basically people will throw a lot of grief at it and say oh it's real heavy. 96 00:12:36,200 --> 00:12:43,440 It's slow. Just use requests. And I agree that if you're consuming from an API then 97 00:12:43,440 --> 00:12:50,320 yeah you know go with requests or if it's a simple web page go with requests. But often 98 00:12:50,320 --> 00:12:59,920 you know we're not dealing with simple web pages here. Right. As you can see we've got 99 00:12:59,920 --> 00:13:07,560 the loading sign. If you actually inspect the source code that they shoot over at you 100 00:13:07,560 --> 00:13:14,280 you'll see that you're working with an I frame. So there's just a lot of JavaScript that's 101 00:13:14,280 --> 00:13:23,000 making this page happen. So you can't just use requests. But that's OK because with the 102 00:13:23,000 --> 00:13:29,000 Selenium authors have said is there's so many tasks that businesses are doing on a day to 103 00:13:29,000 --> 00:13:36,320 day basis that can and should be automated. So if you find yourself you know clicking 104 00:13:36,320 --> 00:13:43,080 through Chrome and you do the same clicks over and over and over again that can be automated. 105 00:13:43,080 --> 00:13:49,480 Right. As I was saying last week you know I'm half me and half Python. So you know if 106 00:13:49,480 --> 00:14:02,160 I want to use Chrome with my Python appendages then you know so be it. And in fact I didn't 107 00:14:02,160 --> 00:14:10,600 really realize it but you know Selenium has been one of my like best used tools in the 108 00:14:10,600 --> 00:14:16,160 past. Right. So out of all the code that I've written you know some of the scripts that 109 00:14:16,160 --> 00:14:21,280 ran the most I mean you know I have written scripts that have run like in the millions 110 00:14:21,280 --> 00:14:34,880 of times and some of those leverage Selenium. It's a great great tool. So. So anywho let's 111 00:14:34,880 --> 00:14:42,400 just go ahead and get down to the brass tacks of this. So we'll basically be using Selenium 112 00:14:42,400 --> 00:14:49,640 plus PDF plumber which I've realized and that's why I said call a plumber. But PDF plumber 113 00:14:49,640 --> 00:14:59,920 is actually just a wrapper around PDF miner. But anywho the code is only so interesting 114 00:14:59,920 --> 00:15:11,360 to so many people but the data is interesting to many. So what we can do here is we can 115 00:15:11,360 --> 00:15:24,320 first read in the PDF. So we've now got this PDF file in memory. And as I was saying you 116 00:15:24,320 --> 00:15:31,800 know you can do a whole bunch of parsing right. So I was originally just trying to parse out 117 00:15:31,800 --> 00:15:42,780 the tables but basically what you can cleverly do here is just iterate over all the images 118 00:15:42,780 --> 00:15:54,840 in the sheet and you basically just stop if one of them gets decoded as a QR code. I saw 119 00:15:54,840 --> 00:16:05,360 originally I was trying to like specify what index the image was but it's basically all 120 00:16:05,360 --> 00:16:16,880 the laboratories like to have their own unique certificate of analysis. So SC labs may have 121 00:16:16,880 --> 00:16:20,620 a different COA I imagine they would. I don't actually know all the different California 122 00:16:20,620 --> 00:16:30,400 laboratories but they all love their COAs and they're real particular about them. There's 123 00:16:30,400 --> 00:16:35,520 a certain compliance requirement as well in California what you can and cannot put on 124 00:16:35,520 --> 00:16:49,640 there. Exactly and so long story short is for any that put the QR code on them we can 125 00:16:49,640 --> 00:16:58,840 utilize that. I wonder how common that is. And that's sort of a downside. So I do know 126 00:16:58,840 --> 00:17:04,200 what so here see they're using Confident Cannabis. I wonder you know what essentially Confident 127 00:17:04,200 --> 00:17:10,080 Cannabis is you know market share is of laboratory tests. That was something Canalytics was always 128 00:17:10,080 --> 00:17:17,180 wondering about. But I think Canalytics can almost instead of just stepping on other people's 129 00:17:17,180 --> 00:17:21,800 toes I find it's more interesting to just solve different pieces of the puzzle. And 130 00:17:21,800 --> 00:17:28,320 so you know they're good at providing certificates right and they've got them online. And so 131 00:17:28,320 --> 00:17:33,760 the piece of the puzzle that I saw that we could fix was okay we can now actually just 132 00:17:33,760 --> 00:17:42,640 wrangle all this data and get it to the consumer and not just Confident Cannabis but also wrote 133 00:17:42,640 --> 00:17:53,480 a thanks to John for sharing the PDFs. I wrote a routine to parse tag leaf limbs COAs. And 134 00:17:53,480 --> 00:18:02,200 so basically the task now is to have this COA parser out there and we basically just 135 00:18:02,200 --> 00:18:10,000 need to identify you know all the COAs that have QR codes on them and then make sure that 136 00:18:10,000 --> 00:18:20,240 we can then parse those results from the web. And then if the lab doesn't have a QR code 137 00:18:20,240 --> 00:18:29,360 at that point we'll have to use the PDF plumber to plumb the PDF. And so that's going to be 138 00:18:29,360 --> 00:18:35,920 a time intensive task. And so this is where I was telling you that I was going to share 139 00:18:35,920 --> 00:18:43,000 with you how I intend to make money with Canlytics and that way you can make money in the exact 140 00:18:43,000 --> 00:18:53,340 same way. And so I've written a and I'll we'll get into the guts of it today because that's 141 00:18:53,340 --> 00:19:01,520 kind of what you would you come here today to see. Right I get to show you all the special 142 00:19:01,520 --> 00:19:06,160 things right we get up we get to pull up the hood and play with the engine and do all the 143 00:19:06,160 --> 00:19:14,840 fun stuff. But that's the point of the cannabis data science to actually get our hands on 144 00:19:14,840 --> 00:19:21,200 this fun stuff. So long story short is we've now got this tool COA doc. And so the idea 145 00:19:21,200 --> 00:19:36,200 is input any arbitrary any arbitrary COA and then parse it. And just to to actually just 146 00:19:36,200 --> 00:19:42,860 show you your maybe just a demonstration of the tool will be better than than showing 147 00:19:42,860 --> 00:19:48,000 you the code. So I'll demonstrate the code I mean I'll demonstrate the tool and then 148 00:19:48,000 --> 00:19:54,600 we can show you the code behind it. So first things first let's just make sure that we 149 00:19:54,600 --> 00:20:03,880 can in fact get the QR code because that's a pretty important part of the whole whole 150 00:20:03,880 --> 00:20:15,280 process. So yes we can get the QR code and then to sprinkle in data science right and 151 00:20:15,280 --> 00:20:23,040 that's actually sort of one of the lessons of the day is a little bit can go a long way 152 00:20:23,040 --> 00:20:31,040 and a little bit of clever you know machine learning clever you know these clever algorithms 153 00:20:31,040 --> 00:20:38,320 can go a long way. And so we only need one line of magic to make a lot happen right. 154 00:20:38,320 --> 00:20:53,800 And so want to give another shout out to Pi Z bar and they're working on cameras and computer 155 00:20:53,800 --> 00:21:03,120 vision to real cool applications. But basically what you can do is you can decode this object. 156 00:21:03,120 --> 00:21:13,280 So decoding and then all of a sudden we've got the decoded QR code data and as you'll 157 00:21:13,280 --> 00:21:24,840 see we've got the URL nice nicely in there. So you know the QR code of course was specifically 158 00:21:24,840 --> 00:21:30,640 designed for computers right. So if you're going to do any computer vision task right 159 00:21:30,640 --> 00:21:35,400 QR code is of course the simplest but that's the way can let it likes to do it right. You 160 00:21:35,400 --> 00:21:43,160 start with the absolute simplest possible use case and then we can take it from there 161 00:21:43,160 --> 00:21:52,520 right before we start like you know applying computer vision to the sample images themselves 162 00:21:52,520 --> 00:21:59,240 which I think we could and maybe should do eventually we can just start simple. So long 163 00:21:59,240 --> 00:22:07,000 story short we can get that QR code just going to fire up the tool real quick and then we 164 00:22:07,000 --> 00:22:21,880 can basically see it in use. So basically I just put this behind a super simple API 165 00:22:21,880 --> 00:22:35,920 just to just show you the actual guts of it. So there's nothing super fancy about the API 166 00:22:35,920 --> 00:22:45,480 itself. So if you make a request it basically tries to find out if you posted any URLs. 167 00:22:45,480 --> 00:22:51,880 So the idea is you can use this programmatically so if you have your own QR code reader in 168 00:22:51,880 --> 00:22:58,440 your app then all you have to do is implement a QR code reader in your app and you can just 169 00:22:58,440 --> 00:23:09,360 scan one of these packages and then post that URL to this endpoint or you can use the Camletics 170 00:23:09,360 --> 00:23:20,360 website and you can upload the files directly. So still working on this so I'm still going 171 00:23:20,360 --> 00:23:27,200 to have to add all the bells and whistles for you so I'm just going to show you under 172 00:23:27,200 --> 00:23:36,240 the hood today so you know so bear with me. But this is the tool COADOC and the idea is 173 00:23:36,240 --> 00:23:50,040 to just be able to upload any arbitrary PDF or scan a QR code and then get that data. 174 00:23:50,040 --> 00:23:58,480 And then under the hood literally three lines of code. I mean of course you have to do all 175 00:23:58,480 --> 00:24:06,080 the rigmarole like check if. I'm doing things like making sure that people are only posting 176 00:24:06,080 --> 00:24:12,560 PDFs and Zips and there's still a bit more safety checks I want to do before I publish 177 00:24:12,560 --> 00:24:20,760 this but essentially you can get it down to just fire up your COA parser you know tell 178 00:24:20,760 --> 00:24:29,880 it to parse your URLs and then you've got your data and keep in mind that the magic 179 00:24:29,880 --> 00:24:39,680 happening is we've basically got two different algorithms here. So I've written a custom 180 00:24:39,680 --> 00:24:48,160 algorithm for confident cannabis and I've written a custom algorithm for tag leaf limbs. 181 00:24:48,160 --> 00:24:55,000 I don't know how prevalent tag leaf limbs is but I see confident cannabis around so 182 00:24:55,000 --> 00:25:02,800 for me I thought it was worthwhile and so the idea is if you know other limbs or you 183 00:25:02,800 --> 00:25:09,280 know other labs you could basically go and make them a value proposition and say hey 184 00:25:09,280 --> 00:25:19,340 I'll write you a parsing algorithm for your PDF that way it can be parsed along with any 185 00:25:19,340 --> 00:25:29,440 other COA by COA doc and just to just put you know just to you know actually put this 186 00:25:29,440 --> 00:25:47,520 to good use here let's just go ahead and drop this classic Jack COA and hopefully 187 00:25:47,520 --> 00:26:09,760 hopefully going to show you things so in production it won't actually open a browser but just 188 00:26:09,760 --> 00:26:16,720 for demonstration hopefully we yep so we just open up this browser look I always say this 189 00:26:16,720 --> 00:26:23,640 is my joke like look mom no hands so just opened up Chrome just got all the heavy metal 190 00:26:23,640 --> 00:26:30,160 results got all the other results hopefully we don't run into errors we may have hit an 191 00:26:30,160 --> 00:26:46,040 error oh no look no hands so so bada bing bada boom and I'll zoom in on this but basically 192 00:26:46,040 --> 00:26:55,120 there we have it so now all you all you have to do is get you know the URL or the actual 193 00:26:55,120 --> 00:27:03,040 COA and as we've seen you know the producers and processors are making an effort to get 194 00:27:03,040 --> 00:27:09,200 those on their packaging so I applaud them for that right so you can have the glass half 195 00:27:09,200 --> 00:27:16,240 full or the glass half empty mentality the glass half empty mentality which I am susceptible 196 00:27:16,240 --> 00:27:24,080 to would be hey you know why aren't why aren't these producers giving us all this data you 197 00:27:24,080 --> 00:27:28,440 know why why don't they have an API you know why can't I just ping this and get all this 198 00:27:28,440 --> 00:27:33,120 data and it's like okay okay you know just you know quit being so demanding right they've 199 00:27:33,120 --> 00:27:39,080 done a pretty good job right they've grown some fantastic cannabis they've got it packaged 200 00:27:39,080 --> 00:27:44,560 up and they've even you know got it nice and labeled and put a QR code on it so it's like 201 00:27:44,560 --> 00:27:49,480 you know they've you know they've taken the ball as far as they can they you know they've 202 00:27:49,480 --> 00:27:54,800 taken as far as they can and it's basically just like okay like thank you you know like 203 00:27:54,800 --> 00:28:01,440 cantlydx will take it the last the last the last yard you know that like I mean think 204 00:28:01,440 --> 00:28:06,080 about all the work that's been done to get the data this far I mean think about this 205 00:28:06,080 --> 00:28:13,160 this is seven analyses here this is an analyst who looked at this under a microscope for 206 00:28:13,160 --> 00:28:21,280 foreign matter they tested it on an ICP for heavy metals not an easy feat they performed 207 00:28:21,280 --> 00:28:30,760 a microbial test a micro toxin test those two tests may take up to 48 hours or so so 208 00:28:30,760 --> 00:28:39,120 those can be timely a pesticide test expensive if you want to go get your sample tested for 209 00:28:39,120 --> 00:28:45,680 pesticides that may cost you north of 300 bucks in California I think good pesticide 210 00:28:45,680 --> 00:28:53,420 tests may cost you almost $800 or so water activity fairly and moisture are fairly simple 211 00:28:53,420 --> 00:28:59,000 tests I think you could if you're clever you could conduct these tests at home probably 212 00:28:59,000 --> 00:29:03,680 with your microwave and a small sample of your flowers so these are pretty easy tests 213 00:29:03,680 --> 00:29:08,440 well maybe not you could do moisture I don't know about water activity but fairly easy 214 00:29:08,440 --> 00:29:13,680 tests but they're still informative so you've got you know your tests here you know you 215 00:29:13,680 --> 00:29:18,360 now know the laboratory if you want to give them a call or you know you want to shoot 216 00:29:18,360 --> 00:29:25,480 them an email we actually didn't get their email in this case but you could go you know 217 00:29:25,480 --> 00:29:29,880 do a search for the laboratory so that way you actually know what laboratory tested your 218 00:29:29,880 --> 00:29:35,280 product just in case you wanted to give them a call you know that of course yeah I mean 219 00:29:35,280 --> 00:29:44,560 you were expecting it to have passed pesticides but check this out we now have 78 compounds 220 00:29:44,560 --> 00:29:52,460 so yes you know they're reporting the THCA and Delta 9 but check this out you know this 221 00:29:52,460 --> 00:29:59,640 wasn't on the label you know the CBGA and once again you can't fault them they only 222 00:29:59,640 --> 00:30:11,200 have so many square inches here right but I think this is an important one like whoa 223 00:30:11,200 --> 00:30:17,760 like look at look at all that CBGA that's a in my opinion an enormous amount of CBGA 224 00:30:17,760 --> 00:30:26,960 and so you know now I know when I consume this cannabis you know in this package there's 225 00:30:26,960 --> 00:30:35,440 almost a hundred gram a hundred milligrams right because this package is 3.5 grams so 226 00:30:35,440 --> 00:30:54,360 now we can do 27.52 times 3.5 whoops there's just shy of a hundred milligrams of CBGA in 227 00:30:54,360 --> 00:31:02,120 this package which is if you're not measuring it you're not managing it so if you know I 228 00:31:02,120 --> 00:31:07,200 like Jack and someone else doesn't like Jack maybe it's got something to do with the CBGA 229 00:31:07,200 --> 00:31:13,400 or maybe maybe it's got to do with one of these many many terpenes in your product right 230 00:31:13,400 --> 00:31:20,160 and so these are what consumers are kind of realizing that they want to be measuring right 231 00:31:20,160 --> 00:31:27,600 so check this out right you've got the terpenolene and then John want to applaud you right so 232 00:31:27,600 --> 00:31:35,200 once again now that I've got my date my great this is my data right because it's a product 233 00:31:35,200 --> 00:31:44,600 that I bought and you know I went and wrangled the data so you know as a consumer right I 234 00:31:44,600 --> 00:31:51,040 think you should be collecting your data but check this out you know let's say I want to 235 00:31:51,040 --> 00:32:04,920 keep track of my beta pionene to d-lemonene ratio beta pionene 0.06 limonene 0.04 which 236 00:32:04,920 --> 00:32:13,720 is a shockingly small amount of limonene you normally see I don't know what the distribution 237 00:32:13,720 --> 00:32:17,920 is off the top of my head but I want to say you normally see more limonene than that in 238 00:32:17,920 --> 00:32:26,160 your products so I don't know if I just you know I kind of like this strains sometimes 239 00:32:26,160 --> 00:32:34,240 I think sometimes I may prefer something else but so am I just maybe adverse to limonene 240 00:32:34,240 --> 00:32:39,800 or maybe I just like the beta pionene or maybe there's something else going on here but you 241 00:32:39,800 --> 00:32:47,840 know it's got terpenelein doesn't it 0.06 over 0.04 it's a high beta pionene strain by 242 00:32:47,840 --> 00:32:55,480 those even even with the small amounts exactly and in fact I this is what I started to figure 243 00:32:55,480 --> 00:33:01,400 out that I may like about Jack is I think the terpenelein at least gives the Jack its 244 00:33:01,400 --> 00:33:08,560 distinctive smell it's just to be frank you know it's a kind of a pungent that people 245 00:33:08,560 --> 00:33:19,520 call it like a almost like a chemically type it's a turpentine smell I got to get when 246 00:33:19,520 --> 00:33:23,840 I first realized this what I got to give him a shout out because they were the ones who 247 00:33:23,840 --> 00:33:29,300 helped me connect the dots it was planet 13 in Las Vegas they were actually labeling the 248 00:33:29,300 --> 00:33:36,560 percentages of terpenes on the products and that was when I mentally connected the dots 249 00:33:36,560 --> 00:33:42,800 between oh you know the reason I like the Jack type strains is the terpenelein and it 250 00:33:42,800 --> 00:33:48,840 was sort of a revolutionary point because it's like oh I don't have to get Jack Harrier 251 00:33:48,840 --> 00:33:55,400 every single time as long as you know it's got some terpenelein in it you know it may 252 00:33:55,400 --> 00:34:01,920 have a you know a similar effect and in fact you can kind of branch out because maybe maybe 253 00:34:01,920 --> 00:34:09,320 some of these you can get something with terpenelein plus something else so in that was sort of 254 00:34:09,320 --> 00:34:15,920 no Keegan I probably should again chime in here on trippin lean is part of a cluster 255 00:34:15,920 --> 00:34:22,520 so if it's going to have terpenelein it's going to have a series of others as well caring 256 00:34:22,520 --> 00:34:29,320 phalaenadrine alpha terpene and gamma trip they run off the same enzyme so while the 257 00:34:29,320 --> 00:34:36,040 terpenelein is the dominant it's the lead of that enzyme the enzyme makes these others 258 00:34:36,040 --> 00:34:42,360 so if you're in a high terpenelein strain you've got these others come along for the 259 00:34:42,360 --> 00:34:45,640 ride as it were. 260 00:34:45,640 --> 00:34:52,440 I love it and now we can finally you know collect data on it and well I guess I guess 261 00:34:52,440 --> 00:34:56,720 the data has already been out there so I guess you already you already know this right so 262 00:34:56,720 --> 00:35:01,600 that's what yeah I've been following this for years it's just what is the utility or 263 00:35:01,600 --> 00:35:09,200 what is the effect I mean is it the terpenelein that's driving the effects in the high terpenelein 264 00:35:09,200 --> 00:35:14,160 strains or is it one of the other terpenes or is it another compound that is also comes 265 00:35:14,160 --> 00:35:24,240 with it we don't know yet but we can only look at what is reported right and this is 266 00:35:24,240 --> 00:35:28,280 this is why I was thinking this was sort of the missing piece of the puzzle right you've 267 00:35:28,280 --> 00:35:36,440 got the lab result data you just need to get this in the hands of the consumer or in your 268 00:35:36,440 --> 00:35:43,520 case you know your research participants that way you know one you know they can act upon 269 00:35:43,520 --> 00:35:48,480 their data and then two you could actually you know match that and study this because 270 00:35:48,480 --> 00:35:56,480 now if you just tell your consumers you know if you and this is why I was saying you know 271 00:35:56,480 --> 00:36:03,080 economic forces now right so if I do this if you do this a bunch of people do this then 272 00:36:03,080 --> 00:36:08,800 it'll put the economic incentive to put your QR code on your package because what you could 273 00:36:08,800 --> 00:36:16,360 say is okay make sure if you're participating in my study make sure to go and get something 274 00:36:16,360 --> 00:36:23,400 with your QR code and scan it you could either have your own app or they maybe they could 275 00:36:23,400 --> 00:36:30,640 just send you the URLs or right and that's the cool thing about this tool is you can 276 00:36:30,640 --> 00:36:36,440 package it up and use it however you want but basically now the consumer doesn't even 277 00:36:36,440 --> 00:36:40,400 have to do anything they just have to buy their package well they do have to they scan 278 00:36:40,400 --> 00:36:47,680 it just beep and then then they it's connected that was that was the only thing that was 279 00:36:47,680 --> 00:36:56,080 missing we've got the lab results the consumer bought it there's just no no match there and 280 00:36:56,080 --> 00:37:03,360 so now if you've got you know your consumer app or your your study app you can now finally 281 00:37:03,360 --> 00:37:12,000 make this connection the consumer you know like myself you know they can get their data 282 00:37:12,000 --> 00:37:21,040 and um thought of being about a boom like I I could see so where are you so what this 283 00:37:21,040 --> 00:37:31,960 is you sucked it up from a QR code that was on the package or was in the PDF or I mean 284 00:37:31,960 --> 00:37:38,920 what you just ran was off of QR code input right and so that's what's so cool about the 285 00:37:38,920 --> 00:37:46,920 algorithm is basically what it all funnels to the URL so it's um and so that's that's 286 00:37:46,920 --> 00:37:51,720 kind of why I wanted to show you under the the hood today so if you look at the parsing 287 00:37:51,720 --> 00:38:03,800 routine okay now I understand basically it just says oh you know so long story short 288 00:38:03,800 --> 00:38:14,640 is the PDF routine subsequently calls the parse URL routine so basically if you have 289 00:38:14,640 --> 00:38:23,000 the the PDF say you're a producer one would hope the producer actually has the raw data 290 00:38:23,000 --> 00:38:29,760 but if for whatever reason they don't then they can just they can just upload this PDF 291 00:38:29,760 --> 00:38:37,720 and it will find this URL from the QR code or if you're just a consumer you can just 292 00:38:37,720 --> 00:38:46,480 scan the QR code and it would just just use this URL directly and I still need to to wire 293 00:38:46,480 --> 00:38:53,560 up this functionality but the idea is you could just however you get that URL that's 294 00:38:53,560 --> 00:39:03,720 sort of up to you so whether you just copy and paste it um 295 00:39:03,720 --> 00:39:11,160 so I'm wondering um of the five PDFs I sent you the other day I wonder I didn't look at 296 00:39:11,160 --> 00:39:21,560 them for URLs for QR codes um I wonder how prevalent that is so with the tag leaf limbs 297 00:39:21,560 --> 00:39:35,000 I was able to get five of six so five so five oh wait a minute tag leaf is a lab um yes 298 00:39:35,000 --> 00:39:46,720 so so here and because I think I sent you several different labs PDFs um here I'll 299 00:39:46,720 --> 00:39:53,880 show you um we've got a little time here um so there's a little glare on my screen okay 300 00:39:53,880 --> 00:40:08,080 so so here um if you're okay I with me using this one then I'll I can demo the tag leaf 301 00:40:08,080 --> 00:40:15,000 limb limbs one um basically let's just talk was that one of the ones I sent you I don't 302 00:40:15,000 --> 00:40:23,920 remember the lab names um yes okay um no you whatever I send you you can use don't worry 303 00:40:23,920 --> 00:40:35,400 about it cool so here let's just just demo this um this sunbeam one real quick then okay 304 00:40:35,400 --> 00:40:44,480 so the idea is okay so we've got the PDF so hopefully we can just get this URL straight 305 00:40:44,480 --> 00:40:54,880 from the QR code so basically you know we're just looking at all the images on the first 306 00:40:54,880 --> 00:41:03,760 page of that one and we should hopefully well first let's just see if we've got an object 307 00:41:03,760 --> 00:41:09,960 so let's see we've found the QR code and so this one is tag leaf limbs and this is where 308 00:41:09,960 --> 00:41:15,880 I was saying like economic forces are just so remarkable because it's basically like 309 00:41:15,880 --> 00:41:23,120 you know once one player introduces the QR code you know everyone else kind of has to 310 00:41:23,120 --> 00:41:29,200 to follow suit because that's now the standard to now tag leaf limbs right they've got a 311 00:41:29,200 --> 00:41:34,280 QR code linking to their certificate online and this isn't actually that hard right when 312 00:41:34,280 --> 00:41:40,920 I worked at the laboratory I implemented this exact same thing you know I uh I'm not bragging 313 00:41:40,920 --> 00:41:47,840 I just think uh you know this is a standard that can be implemented right so if you see 314 00:41:47,840 --> 00:41:52,940 your laboratory and they don't have this that's why I was saying you know you should reach 315 00:41:52,940 --> 00:42:01,120 out to them and don't blame them be the glass half full guy say hey I can meet you or girl 316 00:42:01,120 --> 00:42:05,680 you know say I can be the missing piece of your puzzle right I can meet you where you 317 00:42:05,680 --> 00:42:12,680 are maybe you need help implementing a QR code maybe you need help with an API right 318 00:42:12,680 --> 00:42:21,280 so some labs need more help than others right some limbs need more help than others but 319 00:42:21,280 --> 00:42:34,160 so once again different right so different web page but I find with here I use beautiful 320 00:42:34,160 --> 00:42:44,920 soup to in request so I didn't need selenium to parse this one but I find scraping or parsing 321 00:42:44,920 --> 00:42:59,160 web pages to be easier than parsing PDFs because people use like CSS and things right so just 322 00:42:59,160 --> 00:43:05,680 the the syntax that people use makes it real easy to parse these they're not easy but it's 323 00:43:05,680 --> 00:43:13,400 about a data it's and so actually I shouldn't talk about prices right because technically 324 00:43:13,400 --> 00:43:18,240 you get in trouble you know allowed to just talk about prices really nilly but but long 325 00:43:18,240 --> 00:43:25,560 story short is you know I assess how much time it would take me to parse one of these 326 00:43:25,560 --> 00:43:32,320 and charge accordingly so basically you know if there was a new lens that wanted to a parsing 327 00:43:32,320 --> 00:43:37,880 routine and it doesn't have to be this one and this is what's cool right coa doc is open 328 00:43:37,880 --> 00:43:49,000 source so you can clone this and right you could have your own private and sorry there's 329 00:43:49,000 --> 00:43:58,800 just such a clear here you could have your own private coa doc code right you could just 330 00:43:58,800 --> 00:44:04,360 clone the repository make your own additions or if you wanted to you know benefit the whole 331 00:44:04,360 --> 00:44:12,680 cannabis community you know the whole cannabis industry you could you know make a pull request 332 00:44:12,680 --> 00:44:19,280 you know with your addition right so if you added new limbs you can make a pull request 333 00:44:19,280 --> 00:44:27,440 and then I can review your code if it's valid I could you know incorporate your code here 334 00:44:27,440 --> 00:44:34,760 on the main can lytics website you could get paid for your work coding I would like that 335 00:44:34,760 --> 00:44:42,200 right because it adds to the can lytics code base the laboratory wins because they get 336 00:44:42,200 --> 00:44:51,680 to get they get to get incorporated so long story short it's just it's simply a tool that 337 00:44:51,680 --> 00:45:04,000 can be used to go from URL to data and just yes for one last demo we can just parse this 338 00:45:04,000 --> 00:45:12,720 other strain I got some gmo pie so there's still some work to be done so essentially 339 00:45:12,720 --> 00:45:20,400 if and John this is why I wouldn't mind getting this tool in your hands because essentially 340 00:45:20,400 --> 00:45:26,960 if you've got like a big body of coas that need to be parsed then it'd be cool to just 341 00:45:26,960 --> 00:45:35,640 run it through this see which ones you can parse and then which ones you can't and the 342 00:45:35,640 --> 00:45:41,280 ones you can't maybe they're they're they're on a different lens or a different lab that 343 00:45:41,280 --> 00:45:48,520 has implemented a QR code in which case we can try to write you know like web scraping 344 00:45:48,520 --> 00:45:56,800 routine and then like I said worse comes to worse and they just have a PDF you can use 345 00:45:56,800 --> 00:46:05,320 PDF plumber to parse the tables it's just I would just do that at a at a last case was 346 00:46:05,320 --> 00:46:14,840 like a last resort I would do that at last resort but but as I said it wouldn't be like 347 00:46:14,840 --> 00:46:19,440 I think it would still be worthwhile well worthwhile depends on the costs but it would 348 00:46:19,440 --> 00:46:27,360 still add value to create a parsing routine for confident cannabis PDFs because then you 349 00:46:27,360 --> 00:46:34,240 could parse them offline but I think the marginal benefit of that small the marginal cost is 350 00:46:34,240 --> 00:46:41,640 high I think you're almost always going to have an internet connection and if you don't 351 00:46:41,640 --> 00:46:49,160 then you've got bigger problems than parsing than parsing COAs and so long so long story 352 00:46:49,160 --> 00:46:55,920 short is this tool works pretty well if you've got an internet connection and then you can 353 00:46:55,920 --> 00:47:02,120 get a you can get all these results and in there's still work to be done so for example 354 00:47:02,120 --> 00:47:06,920 like a minor thing that needs to be done is these results need to be cleaned up a little 355 00:47:06,920 --> 00:47:13,840 bit but that that's minor so there's still some polishing that needs to be done so stay 356 00:47:13,840 --> 00:47:29,000 tuned before I you know formally publish everything huh okay um I think um there are from my perspective 357 00:47:29,000 --> 00:47:37,680 two folks two constituencies to be talking to and one of them I'm about to have a conversation 358 00:47:37,680 --> 00:47:45,320 with um is the lab that I work very closely with they the scientific so I'll check what 359 00:47:45,320 --> 00:47:52,360 their COA publication status is if they've got QR codes on there or not so that you know 360 00:47:52,360 --> 00:47:59,400 I mean would be good great to have make sure VEDA can have their um their sets their their 361 00:47:59,400 --> 00:48:07,400 data sets read by uh folks like us um the second one is is a retail store like floor 362 00:48:07,400 --> 00:48:13,720 and they're the ones who have corpuses of COAs so it might be useful uh what I sent 363 00:48:13,720 --> 00:48:18,920 you was the five that they would say sent me for a couple strings we were evaluating 364 00:48:18,920 --> 00:48:25,360 but they have about 70 or so um I'll be having a conversation with them and we can see what 365 00:48:25,360 --> 00:48:33,440 their corpus of COAs are and how readily they could be sucked up into a data set because 366 00:48:33,440 --> 00:48:40,160 that's where the real utility or a real utility of this is is because then in a store can 367 00:48:40,160 --> 00:48:46,680 decide how their inventory parses or you know how to display it how to categorize it or 368 00:48:46,680 --> 00:48:52,800 all that right now they just you know we're teaching them to read COAs but there's no 369 00:48:52,800 --> 00:48:58,840 automation to it you know you it's tough to run the algorithms that we want to run on 370 00:48:58,840 --> 00:49:06,000 so um I would say you know you we go lab by lab or you know labs that are affiliated with 371 00:49:06,000 --> 00:49:11,400 the efforts we're doing and then with the retail as well and um yeah it's a good start 372 00:49:11,400 --> 00:49:18,600 I love it John and let's bring let's bring VEDA on board because that's the whole idea 373 00:49:18,600 --> 00:49:26,640 is just to keep incorporating labs and limbs until we can parse you know the majority of 374 00:49:26,640 --> 00:49:35,200 COAs right so like if we can parse 80% of COAs that would be phenomenal yeah and like 375 00:49:35,200 --> 00:49:43,880 you said right the producer has access to their their data and their client portal but 376 00:49:43,880 --> 00:49:50,560 the retailer the retailers just getting sometimes I mean probably hundreds if not thousands 377 00:49:50,560 --> 00:49:56,800 of COAs from all sorts of different vendors and so if now we just have this nice standardized 378 00:49:56,800 --> 00:50:04,640 tool where they could just take a folder of COAs from five different labs right or and 379 00:50:04,640 --> 00:50:09,040 they they don't even have to think about what limbs it comes from they just right right 380 00:50:09,040 --> 00:50:18,040 they just drag and drop it parses and then they get their data in in in my opinion it's 381 00:50:18,040 --> 00:50:26,600 their data right the the lab testing in my opinion is two products one it's the it's 382 00:50:26,600 --> 00:50:36,200 the signature essentially the the certificate right that yes this was certified tested is 383 00:50:36,200 --> 00:50:42,160 the data in this is mission critical right this needs right because this is for menu 384 00:50:42,160 --> 00:50:50,360 boards this is for labels this is for getting in consumers hands and we're just meeting 385 00:50:50,360 --> 00:50:54,960 the industry where it is right as I said it would be phenomenal right if every single 386 00:50:54,960 --> 00:51:02,040 company had has implemented an API and QR codes and this and this and this but right 387 00:51:02,040 --> 00:51:08,840 they've got a group cannabis and they've got to sell cannabis and test cannabis right people's 388 00:51:08,840 --> 00:51:14,680 got so many things to do then that's why you know the campus data science team is here 389 00:51:14,680 --> 00:51:20,520 to fill in this last piece of the puzzle right so we'll meet you where you are thank you 390 00:51:20,520 --> 00:51:26,400 right it's like thank you Tackleaf Limbs for publishing your results and you know thank 391 00:51:26,400 --> 00:51:31,040 you Confident Cannabis and thank you Cali Green Laboratory and everyone else you know 392 00:51:31,040 --> 00:51:37,240 thank you for doing your part and then we're basically we'll do our part and so basically 393 00:51:37,240 --> 00:51:47,520 our parts just just get everything get all the data and standardize it and then you know 394 00:51:47,520 --> 00:51:57,520 put it nicely in your hands for you to use so so you've got to see the first unveiling 395 00:51:57,520 --> 00:52:06,680 of COA doc you basically got to see the complete raw under the hood unfinished version so today 396 00:52:06,680 --> 00:52:12,440 I'll put all the bells and whistles on it but now you know how the magic happens right 397 00:52:12,440 --> 00:52:21,040 so now when you simply drag your COAs here and get your nice crispy clean data you know 398 00:52:21,040 --> 00:52:29,040 how it's working under the hood and it's all open source and I think that's what's 399 00:52:29,040 --> 00:52:35,560 going to make this tool awesome right because I don't think I'm the first one who's written 400 00:52:35,560 --> 00:52:42,960 in a COA parsing routine in fact I know people who've done this right this is done by a lot 401 00:52:42,960 --> 00:52:47,960 of software companies but if we do it in an open source manner then right like Candace 402 00:52:47,960 --> 00:52:55,320 right you can contribute and get paid for it right you can go to MCR labs in Massachusetts 403 00:52:55,320 --> 00:53:03,420 or who have you and you know see if they want a parsing routine and you know if we kind 404 00:53:03,420 --> 00:53:10,480 of get this tool used by enough people then it would be a no-brainer for the laboratory 405 00:53:10,480 --> 00:53:14,200 or the limbs and they would say yeah like of course we want to be incorporated like 406 00:53:14,200 --> 00:53:21,440 of course we want to get our data to the consumers so got to start somewhere so it's not it's 407 00:53:21,440 --> 00:53:31,480 far from finished but we can at least get the the data for for some samples tested here 408 00:53:31,480 --> 00:53:39,980 in California by Coffin and cannabis so I'll start collecting my results I was kind of 409 00:53:39,980 --> 00:53:45,840 thinking about if I wanted to publish my own personal consumption data that'd be a little 410 00:53:45,840 --> 00:53:56,820 wild but I'm considering it but but anyways I think this would be real interesting data 411 00:53:56,820 --> 00:54:01,860 I'm going to find value out of it so I'm going to start recording my own consumption data 412 00:54:01,860 --> 00:54:09,380 Candace if you're and you don't necessarily have to buy the products right you could potentially 413 00:54:09,380 --> 00:54:19,360 just go to the store if you want to if you want to start recording your consumption data 414 00:54:19,360 --> 00:54:26,200 just use our dosing project app and because the goal is going to be to marry CoA's with 415 00:54:26,200 --> 00:54:35,560 it in the version 2 but if you're building or you're starting to build you know you're 416 00:54:35,560 --> 00:54:41,640 starting to create the tunnel that we're going to be using to feed into the app but the app 417 00:54:41,640 --> 00:54:48,440 will record your use now or you know it can and it asks you dose or how much you're using 418 00:54:48,440 --> 00:54:56,000 and all and it ultimately will calculate dose based or amounts based on CoA's at the moment 419 00:54:56,000 --> 00:55:03,680 it calculates it based on the canonical cannabis class one two or three but at least it's a 420 00:55:03,680 --> 00:55:11,000 recording of your use and so that's the beginnings of it for you so I would recommend you know 421 00:55:11,000 --> 00:55:16,040 trying that out go to the CESC website if you haven't and register you know sign up 422 00:55:16,040 --> 00:55:18,120 for dosing project. 423 00:55:18,120 --> 00:55:22,720 I definitely will because this is something that this is where this is going this is why 424 00:55:22,720 --> 00:55:28,880 I'm asking you to you know kind of help do this is because this is a way to get that 425 00:55:28,880 --> 00:55:36,000 CoA data into dosing project so then we can really get those response curves and all that. 426 00:55:36,000 --> 00:55:41,240 I definitely will because as I would that was sort of my main point today was you know 427 00:55:41,240 --> 00:55:49,480 if you're not measuring it you're not managing it and I'm thrilled I'm just tickled so just 428 00:55:49,480 --> 00:55:55,960 so you know I just pulled this off last night so last night was the first time it you know 429 00:55:55,960 --> 00:56:05,400 it successfully you know I dropped PDF there and it you know came back with the results 430 00:56:05,400 --> 00:56:11,480 so it just happened like late last night and there's still some errors this morning so 431 00:56:11,480 --> 00:56:20,840 luckily the crew demo worked today but I'm like I said I'm tickled with this because 432 00:56:20,840 --> 00:56:28,000 I'm just thrilled because this is something that I've been wanting to keep track of myself 433 00:56:28,000 --> 00:56:34,660 through the longest time and it's just not been possible and as I said there's still 434 00:56:34,660 --> 00:56:41,320 some improvement right now I'm just going to be saving these JSON files but I mean we've 435 00:56:41,320 --> 00:56:50,000 proved that we can work with JSON files so I mean the step one right episode one right 436 00:56:50,000 --> 00:56:56,560 get data right get the data so first things first right before we calculate all the cool 437 00:56:56,560 --> 00:57:05,980 statistics you know we just need to get the actual data but then once we do we can find 438 00:57:05,980 --> 00:57:11,600 all these cool things right so for example here I just calculated a statistic right in 439 00:57:11,600 --> 00:57:25,960 this one eighth of classic Jack there's almost 1.5 grams of THCA which is quite extraordinary 440 00:57:25,960 --> 00:57:35,760 when you think about it so long story short is you know I'll be collecting my data and 441 00:57:35,760 --> 00:57:42,000 you know maybe before long it may be embarrassing but I think I'm going to publish it anyways 442 00:57:42,000 --> 00:57:47,240 just to you know walk the walk and that's what we did today at the cannabis data science 443 00:57:47,240 --> 00:57:54,960 right we walk the walk because this is what can litig set out to do was to get these results 444 00:57:54,960 --> 00:58:03,760 in the consumers hands and we've at least done it we've at least done it once right 445 00:58:03,760 --> 00:58:09,680 if you start on you know start collecting your COAs the way you're doing it and parsing 446 00:58:09,680 --> 00:58:16,760 them and if you start entering your use for those strains in the version of dosing project 447 00:58:16,760 --> 00:58:23,080 which is running I can run charts for you which you probably can't do yet but I can 448 00:58:23,080 --> 00:58:28,960 do pretty easily to start to show you what it would look like in version 2 so that's 449 00:58:28,960 --> 00:58:34,760 a good opportunity I would you know we can talk more about that but why don't you you 450 00:58:34,760 --> 00:58:40,640 know don't publish what you're using do it and I'll show you and then you can make this 451 00:58:40,640 --> 00:58:44,880 exactly because you know I wouldn't encourage anyone to necessarily publish right because 452 00:58:44,880 --> 00:58:51,800 this is great if you're using cannabis medicinally right this that'd be like publishing your 453 00:58:51,800 --> 00:59:00,520 prescription that's sort of an odd thing to do but the idea is right I'm a time series 454 00:59:00,520 --> 00:59:07,160 you know I know a lot of time series statistics so you know I can start to trend to my consumption 455 00:59:07,160 --> 00:59:13,440 over time and like you said John you've got a whole suite of tools that for this this 456 00:59:13,440 --> 00:59:19,640 type of analysis so yeah I mean the version of dosing project which is running now you 457 00:59:19,640 --> 00:59:26,960 know I mean it was designed for a couple of high-level indications for pain and disordered 458 00:59:26,960 --> 00:59:31,960 sleep but that doesn't matter at the moment you know we're gonna make it broader for version 459 00:59:31,960 --> 00:59:39,160 2 but it we're clujing it so you know for the moment just say you know pick one you're 460 00:59:39,160 --> 00:59:46,360 dosing for pain it doesn't matter but what I would recommend well there's a just a four 461 00:59:46,360 --> 00:59:55,040 point scale for how well it worked for you again just don't don't use it in terms of 462 00:59:55,040 --> 01:00:00,840 pain use it in terms of I don't know you know how enjoyable or how you did with it or whatever 463 01:00:00,840 --> 01:00:05,920 there will be mood scales and all that that we're going to incorporate into this but that's 464 01:00:05,920 --> 01:00:16,480 coming in version 2 but you know even we can you know just by you recording your no response 465 01:00:16,480 --> 01:00:22,000 all partial response almost complete or complete and we start mapping that against strains 466 01:00:22,000 --> 01:00:27,640 and we start to get time series which we haven't had and that would be a help this is where 467 01:00:27,640 --> 01:00:33,200 we're going and so you can help prototype it based on coa so yeah go forth and prosper 468 01:00:33,200 --> 01:00:37,800 I think it's great I mean I love your problems figuring out where the app is but it should 469 01:00:37,800 --> 01:00:47,720 be pretty clear from the the CESC website I'll find it today all right because essentially 470 01:00:47,720 --> 01:00:54,000 you're just adding more and more data points right because yeah I mean yeah yeah and you're 471 01:00:54,000 --> 01:01:05,000 adding critical data points right so right you know post consumption surveys so so that 472 01:01:05,000 --> 01:01:10,320 way you could actually you know start to make inference and then if you got like a need 473 01:01:10,320 --> 01:01:18,920 of a diverse panel of participants like your statistics would just be phenomenal and that's 474 01:01:18,920 --> 01:01:25,080 the point that's what is you know that's our initiative and CESC is to do this that's that's 475 01:01:25,080 --> 01:01:30,400 the dosing project in fact we just wrote up a description on will be in the cannabinoid 476 01:01:30,400 --> 01:01:37,080 playbook in August describing our approach with observational studies so if you're not 477 01:01:37,080 --> 01:01:41,320 registered for that I'll send you the link so that you can sign up for the cannabinoid 478 01:01:41,320 --> 01:01:47,920 playbook that comes out each month and just want to thank Candice real quick so Candice 479 01:01:47,920 --> 01:01:54,600 was mentioning that contributing Massachusetts and Florida data and Candice has already helped 480 01:01:54,600 --> 01:02:02,680 with the code for psi labs I just realized that we've written collection routines for 481 01:02:02,680 --> 01:02:12,080 MCR labs and SC labs so I bet you that those two routines could get incorporated so I bet 482 01:02:12,080 --> 01:02:19,840 you with a little bit of work we could read COAs from anyone who does confident cannabis 483 01:02:19,840 --> 01:02:30,040 tag leaf limbs SC labs or MCR labs in Massachusetts so we've still got a long way to go but but 484 01:02:30,040 --> 01:02:35,520 exactly with with the fine folks here of the cannabis data science team we can just get 485 01:02:35,520 --> 01:02:41,920 more and more people on board it's open source so labs could potentially write their own 486 01:02:41,920 --> 01:02:47,520 parsing routines or create their own API so the labs can still do this themselves we're 487 01:02:47,520 --> 01:02:57,560 just we're just filling in that piece of the puzzle for them just to help too cool well 488 01:02:57,560 --> 01:03:02,760 I'll finish this up so that way I can actually get the tool in your hand so thanks for the 489 01:03:02,760 --> 01:03:11,840 demo and your feedback all right but hopefully you found something of value here um I'm 490 01:03:11,840 --> 01:03:16,960 just thrilled I think this is going to be I think it's going to be a groundbreaking 491 01:03:16,960 --> 01:03:22,840 tool so we just need to get it into to more people's hands so that's what I'll be working 492 01:03:22,840 --> 01:03:28,920 on direction that's really great definitely and I want to say couldn't have done it without 493 01:03:28,920 --> 01:03:34,440 both of you couldn't have done it without one you pointing me down this trail John and 494 01:03:34,440 --> 01:03:42,440 also you know providing the initial coas to actually start parsing because as I said I 495 01:03:42,440 --> 01:03:46,560 was originally just going to start parsing the PDF like many people have done in the 496 01:03:46,560 --> 01:03:53,160 past and then I stumbled upon the QR code no that's a good way all righty well cool 497 01:03:53,160 --> 01:04:00,760 cool thanks too cool well thank you both thank you both for advancing cannabis science and 498 01:04:00,760 --> 01:04:09,760 until next week have fun thank you tons of fun very impressed ah I'm impressed with both 499 01:04:09,760 --> 01:04:15,800 of you too so thank you Candice thank you John bye bye until next week see you next 500 01:04:15,800 --> 01:04:31,240 week bye bye now