1
00:00:00,000 --> 00:00:12,400
They're going to make a bunch of TensorFlow announcements.

2
00:00:12,400 --> 00:00:13,920
You are absolutely correct.

3
00:00:13,920 --> 00:00:18,280
So I thought that was kicking off yesterday.

4
00:00:18,280 --> 00:00:24,080
I'd be happy to hear about it if you attended yesterday and then I'm okay with ending a

5
00:00:24,080 --> 00:00:25,560
little early today.

6
00:00:25,560 --> 00:00:28,000
I'm actually in the process of moving.

7
00:00:28,000 --> 00:00:33,080
So I'll pack up and read it.

8
00:00:33,080 --> 00:00:38,160
But Charles, did you attend the Google I.O. yesterday by any chance?

9
00:00:38,160 --> 00:00:40,960
Yes, yeah I did.

10
00:00:40,960 --> 00:00:48,520
You know they showed off their quantum computer lab and they're working on like this conversational

11
00:00:48,520 --> 00:00:53,400
AI where you can converse with objects.

12
00:00:53,400 --> 00:01:00,160
They have a conversation with a paper airplane and with the planet Pluto and stuff.

13
00:01:00,160 --> 00:01:01,160
It was pretty cool.

14
00:01:01,160 --> 00:01:02,160
It was interesting stuff.

15
00:01:02,160 --> 00:01:06,520
I know what I'd ask a paper airplane.

16
00:01:06,520 --> 00:01:12,400
And you said today they're doing stuff with TensorFlow.

17
00:01:12,400 --> 00:01:15,880
I didn't realize Google was behind TensorFlow.

18
00:01:15,880 --> 00:01:23,000
Have you used any machine learning libraries before or anything like that?

19
00:01:23,000 --> 00:01:27,080
Yeah I use TensorFlow all the time.

20
00:01:27,080 --> 00:01:34,960
So I know they made some Firebase announcements yesterday but after like three hours of conference

21
00:01:34,960 --> 00:01:37,680
I can't remember exactly what they were.

22
00:01:37,680 --> 00:01:47,200
They are doing some Firebase presentations today or tomorrow I think.

23
00:01:47,200 --> 00:01:49,960
So I'll have to check in maybe after the fact.

24
00:01:49,960 --> 00:01:54,080
But yeah I'd love to hear about some of the work you're doing with TensorFlow sometime

25
00:01:54,080 --> 00:02:01,640
because it looks like there's good uses but I've never pinpointed down a particularly

26
00:02:01,640 --> 00:02:02,640
good use.

27
00:02:02,640 --> 00:02:06,920
But I haven't tried.

28
00:02:06,920 --> 00:02:19,200
I did a project that located trees for like for a carbon office and for applications like

29
00:02:19,200 --> 00:02:22,120
help to calculate carbon offsets.

30
00:02:22,120 --> 00:02:30,360
And then I did one with the detected like fire extinguishers and AEDs and stairwells

31
00:02:30,360 --> 00:02:38,060
for like for helping people find exits or emergency equipment during some sort of emergency

32
00:02:38,060 --> 00:02:39,640
in high-rise buildings.

33
00:02:39,640 --> 00:02:42,600
Oh that sounds really interesting.

34
00:02:42,600 --> 00:02:45,360
Yeah that was a cool project.

35
00:02:45,360 --> 00:02:52,080
Is that something that so I didn't know if you're are you more like freelance with your

36
00:02:52,080 --> 00:02:56,840
projects or do you work for directly for a company?

37
00:02:56,840 --> 00:02:57,840
I'm a freelancer.

38
00:02:57,840 --> 00:02:58,840
That's cool.

39
00:02:58,840 --> 00:03:03,480
So you get to see a lot of different cool things coming your way.

40
00:03:03,480 --> 00:03:04,480
Yeah.

41
00:03:04,480 --> 00:03:05,480
Let's see.

42
00:03:05,480 --> 00:03:08,880
And then how about you Paul?

43
00:03:08,880 --> 00:03:17,120
I heard from you a little bit this past week and so it sounds like you were making some

44
00:03:17,120 --> 00:03:18,120
progress.

45
00:03:18,120 --> 00:03:19,120
Yeah.

46
00:03:19,120 --> 00:03:20,120
Yeah.

47
00:03:20,120 --> 00:03:22,520
So thanks for that.

48
00:03:22,520 --> 00:03:26,640
I appreciate kind of directing me a little bit there on where to look.

49
00:03:26,640 --> 00:03:30,120
So I took about four different data sets.

50
00:03:30,120 --> 00:03:34,860
I took about 30 or 40,000 records out of each of those.

51
00:03:34,860 --> 00:03:41,680
And then I just threw with my subsets I just threw those into access joined the tables

52
00:03:41,680 --> 00:03:45,400
together per the joins that you were talking about.

53
00:03:45,400 --> 00:03:48,080
And Charles sorry just to fill you in.

54
00:03:48,080 --> 00:03:52,960
So I was looking to do looking at some of the products to get familiar with the different

55
00:03:52,960 --> 00:03:55,560
types of product categories.

56
00:03:55,560 --> 00:03:58,600
Actually I could probably share my screen.

57
00:03:58,600 --> 00:04:02,160
Let's see.

58
00:04:02,160 --> 00:04:08,080
I don't know which I've not used this before.

59
00:04:08,080 --> 00:04:09,080
So let me see.

60
00:04:09,080 --> 00:04:11,280
I can share a tab, a window.

61
00:04:11,280 --> 00:04:15,040
I'll try to share a window here.

62
00:04:15,040 --> 00:04:22,760
I don't know if you can see this.

63
00:04:22,760 --> 00:04:25,560
Let me know if you can see this.

64
00:04:25,560 --> 00:04:26,560
Okay.

65
00:04:26,560 --> 00:04:27,560
Yeah.

66
00:04:27,560 --> 00:04:28,560
Okay.

67
00:04:28,560 --> 00:04:29,560
So very simple.

68
00:04:29,560 --> 00:04:35,720
I thought it would be much quicker if I just took my subsets into access.

69
00:04:35,720 --> 00:04:45,600
So what I was doing was just tying the sales items to inventories and then inventory types

70
00:04:45,600 --> 00:04:49,000
because what I'm interested in looking at is just to get familiar with the different

71
00:04:49,000 --> 00:04:54,280
types of products so I could potentially do my market basket analysis on the different

72
00:04:54,280 --> 00:04:55,460
sales products.

73
00:04:55,460 --> 00:05:04,360
So I joined those together and I don't know if you guys use access much.

74
00:05:04,360 --> 00:05:07,440
I don't use it very often but I just thought this would be a quick and easy way to do some

75
00:05:07,440 --> 00:05:08,720
things.

76
00:05:08,720 --> 00:05:10,000
Just exploratory type stuff.

77
00:05:10,000 --> 00:05:14,600
So you can once you tie your tables together you can drag any of the fields down into this

78
00:05:14,600 --> 00:05:21,800
screen on the bottom and then you can just essentially run the results and look at the

79
00:05:21,800 --> 00:05:24,400
resulting data set.

80
00:05:24,400 --> 00:05:33,600
And so here I've got I just pulled in sales item ID and name and then here I've got the

81
00:05:33,600 --> 00:05:42,980
intermediate product type I guess and then I've got the actual inventory name.

82
00:05:42,980 --> 00:05:47,820
So one of the things that I've noticed on here at least for the work I wouldn't mind

83
00:05:47,820 --> 00:05:54,240
doing is intermediate type is obviously very high level and then you've got these kind

84
00:05:54,240 --> 00:06:01,440
of inventory names but I think these inventory names could actually be subgrouped.

85
00:06:01,440 --> 00:06:07,000
So if I have something called CBD capsules times 30 there might be in here CBD capsules

86
00:06:07,000 --> 00:06:13,040
times 60 which really doesn't it may be too granular.

87
00:06:13,040 --> 00:06:17,480
So maybe there's some other maybe just CBD capsules right.

88
00:06:17,480 --> 00:06:20,840
So I think there might be some subsetting that I might be able to do that would make

89
00:06:20,840 --> 00:06:26,200
my analysis a little more meaningful towards not so granular.

90
00:06:26,200 --> 00:06:31,360
So as we were talking about last time with market basket analysis if you buy peanut butter

91
00:06:31,360 --> 00:06:35,440
and jelly you might buy bread that's kind of the idea.

92
00:06:35,440 --> 00:06:41,200
Well you don't necessarily want to know if you're buying Jif peanut butter or Skippy

93
00:06:41,200 --> 00:06:47,960
peanut butter or you may that may be of interest to somebody but I think for what I want to

94
00:06:47,960 --> 00:06:55,800
do that may be too granular so for CBD capsules times 30 and CBD capsules times 60 that's

95
00:06:55,800 --> 00:06:58,000
probably too granular for what I'm looking at.

96
00:06:58,000 --> 00:07:05,360
So I may end up subsetting these into different groups based on what I see here for part of

97
00:07:05,360 --> 00:07:12,000
the analysis but so that's kind of what I was doing just as a quick and dirty way of

98
00:07:12,000 --> 00:07:18,760
just to get familiar with what's there and I want to continue building out the tables

99
00:07:18,760 --> 00:07:23,920
pulling them in taking a sample and then just pulling into and start creating this kind

100
00:07:23,920 --> 00:07:30,080
of primary key map that I could give you guys and actually give you a whole copy of this

101
00:07:30,080 --> 00:07:37,120
if you're interested in it but to develop this map of the tables because just having

102
00:07:37,120 --> 00:07:42,760
this kind of map of visual representation in one place just gets people oriented to

103
00:07:42,760 --> 00:07:49,280
what's there quickly and I know that you know I'm a visual person so if I see something

104
00:07:49,280 --> 00:07:53,200
like this it's going to stick with them a lot longer if I try to look at a bunch of

105
00:07:53,200 --> 00:08:01,240
different code samples and see where all the relationships are but another thing I was

106
00:08:01,240 --> 00:08:06,520
going to ask both of you this question now with the work that you've done with this data

107
00:08:06,520 --> 00:08:12,880
in the past have you considered maybe taking those large tables and upload it into like

108
00:08:12,880 --> 00:08:18,120
something like Amazon web services or something like that to where you could actually do your

109
00:08:18,120 --> 00:08:28,280
Python development work on your desktop but then tie into the data from a cloud provider.

110
00:08:28,280 --> 00:08:36,200
Good question so I have made some of the lab results available through Firebase actually

111
00:08:36,200 --> 00:08:44,720
like Charles was mentioning so that's basically Google's shot at you know the AWS so that's

112
00:08:44,720 --> 00:08:47,080
Google's version of AWS essentially.

113
00:08:47,080 --> 00:08:48,560
I see okay.

114
00:08:48,560 --> 00:08:55,760
So I think it's possible for some of the smaller data sets and that would be awesome.

115
00:08:55,760 --> 00:09:02,040
Some of the larger ones like sales I think it just may be too costly to put them in the

116
00:09:02,040 --> 00:09:11,320
cloud so I've got a friend David Busby who runs a company OpenTHC and essentially we've

117
00:09:11,320 --> 00:09:21,560
got a new member but essentially what he does is he just stores all the data on his computer

118
00:09:21,560 --> 00:09:28,400
and just runs it as an SQL or SQL access.

119
00:09:28,400 --> 00:09:39,560
Hello Ivan, welcome to the Cannabis Data Science Meetup Group.

120
00:09:39,560 --> 00:09:51,080
I think you're muted Ivan.

121
00:09:51,080 --> 00:09:56,000
Well Ivan we're talking about some leaf data sets and how to access it.

122
00:09:56,000 --> 00:10:06,880
So essentially I can point you in the direction of it but David's you know tried to make some

123
00:10:06,880 --> 00:10:12,040
of the data sort of accessible through SQL queries so mostly like aggregation queries

124
00:10:12,040 --> 00:10:22,960
like totals by day.

125
00:10:22,960 --> 00:10:25,760
Am I still with you guys or am I lagging?

126
00:10:25,760 --> 00:10:33,920
Okay great.

127
00:10:33,920 --> 00:10:38,680
Yeah I mean I thought of like putting it on like a local MySQL database or some other

128
00:10:38,680 --> 00:10:48,520
open database locally but yeah all those cloud services are spendy and this is a lot of data.

129
00:10:48,520 --> 00:10:55,680
It looks like me have lost Paul but I can't explain when he gets back.

130
00:10:55,680 --> 00:11:00,520
What was the second part of his question?

131
00:11:00,520 --> 00:11:05,320
There was the putting it on the cloud but what was the first part of his question do

132
00:11:05,320 --> 00:11:06,320
you remember?

133
00:11:06,320 --> 00:11:16,240
Wasn't it about like accessing it through like SQL or some database queries?

134
00:11:16,240 --> 00:11:23,400
Well I don't know when Paul gets back we'll have to ask him but you know I like the work

135
00:11:23,400 --> 00:11:31,640
he did there by just you know organizing the data in the access so that would be nice to

136
00:11:31,640 --> 00:11:37,760
add the rest of the tables because as we all agree you know the first rule about data science

137
00:11:37,760 --> 00:11:42,640
is you know look at the data.

138
00:11:42,640 --> 00:11:59,040
Well I guess just real quick do you just want to see a like a dummy variable regression

139
00:11:59,040 --> 00:12:05,640
I ran or what are some of the things you're working on Jarvis?

140
00:12:05,640 --> 00:12:17,880
Usually I've been working on this Kaggle competition to classify radio signals as being like extraterrestrial

141
00:12:17,880 --> 00:12:27,560
or just noise like you know like the example they use is like the Voyager like the signals

142
00:12:27,560 --> 00:12:31,640
that they get from the Voyager and like how to detect that from just like space noise.

143
00:12:31,640 --> 00:12:36,840
So I've been kind of spending the last week doing that I haven't really worked on cannabis

144
00:12:36,840 --> 00:12:40,760
data science since last week but I will get back into it.

145
00:12:40,760 --> 00:12:46,200
Although when did you say you were working on this past week again?

146
00:12:46,200 --> 00:12:54,200
It's a Kaggle competition they have it's to classify radio signals as like and you try

147
00:12:54,200 --> 00:13:00,680
and figure out if it's just like space noise or if it's like a if it's like a signal coming

148
00:13:00,680 --> 00:13:07,960
from a space probe or from extraterrestrials or but they use like the Voyager you know

149
00:13:07,960 --> 00:13:12,520
like they like the example they give you like give you example signals from Voyager and

150
00:13:12,520 --> 00:13:19,160
what they look like against space noise and how to filter that signal out.

151
00:13:19,160 --> 00:13:25,640
So do you know all about the spacing on the names but all you know that the tests for

152
00:13:25,640 --> 00:13:34,520
white noise like the stationarity tests is that something that you know about or?

153
00:13:34,520 --> 00:13:40,760
No I'm just kind of exploring it right now I'm just trying to learn stuff.

154
00:13:40,760 --> 00:13:46,840
Well for time series forecasting like some of the forecasting I showed you the the Arema

155
00:13:46,840 --> 00:13:54,080
and the VAR models if you're really doing that rigorously before you start forecasting

156
00:13:54,080 --> 00:14:00,920
you essentially want to make sure your series is what they call stationary so it's essentially

157
00:14:00,920 --> 00:14:09,080
white noise so there's no apparent trends or you know cyclical behavior or anything

158
00:14:09,080 --> 00:14:10,080
like that.

159
00:14:10,080 --> 00:14:17,400
So typically you know when you get the data set and you're going to forecast it you know

160
00:14:17,400 --> 00:14:23,360
you'd often maybe take the natural log and maybe even the first difference to get it

161
00:14:23,360 --> 00:14:31,280
into growth rates and the whole point of that is to take you know when you look at most

162
00:14:31,280 --> 00:14:43,160
series right they're just this trend like so they can you know bias your forecasts so

163
00:14:43,160 --> 00:14:48,840
what you do is you basically just difference it maybe even a couple times and that's why

164
00:14:48,840 --> 00:14:59,520
it's called a ream a reema that's the eye the integration or the difference and so basically

165
00:14:59,520 --> 00:15:05,600
what you're trying to do is you're trying to make the data look like white noise because

166
00:15:05,600 --> 00:15:10,720
it's just a theoretical forecasting and one of the assumptions is that you have white

167
00:15:10,720 --> 00:15:13,920
noise.

168
00:15:13,920 --> 00:15:22,040
So long story short there's a whole bunch of statistical tests you can do to test if

169
00:15:22,040 --> 00:15:29,280
the data is white noise and so that could be useful in your situation.

170
00:15:29,280 --> 00:15:30,280
Thanks.

171
00:15:30,280 --> 00:15:40,240
Hi guys sorry my wife dropped her calculator on her surge detector that turned off our

172
00:15:40,240 --> 00:15:47,560
modem so or not a modem but you know our router so bizarre anyway sorry about that.

173
00:15:47,560 --> 00:15:55,920
We lost you there for a second so but to bounce back to your question so I put some of the

174
00:15:55,920 --> 00:16:03,480
lab results in firebase which is you know google's version of AWS just to sort of make

175
00:16:03,480 --> 00:16:09,560
those accessible through an API I still need to do a bit more work there.

176
00:16:09,560 --> 00:16:15,560
The reason I haven't really done it with sales or inventory is I just think that cost may

177
00:16:15,560 --> 00:16:26,760
be exorbitant and so at OpenTHC what they've done is they basically just store it on like

178
00:16:26,760 --> 00:16:35,360
a server and then they just do like SQL queries to just do aggregates to do daily totals or

179
00:16:35,360 --> 00:16:43,320
your total sales by licensee by money so that's sort of the most accessible the data could

180
00:16:43,320 --> 00:16:51,120
get honestly I think there's potential if you thought of a you know a cost-effective

181
00:16:51,120 --> 00:16:58,640
way to store the data and to essentially serve it up through an API or some other means for

182
00:16:58,640 --> 00:17:01,380
people to consume it.

183
00:17:01,380 --> 00:17:05,560
I think there's a lot of people that would like that data because yes you can do the

184
00:17:05,560 --> 00:17:14,480
public records request but that's you know a pain and I think that that's a barrier to

185
00:17:14,480 --> 00:17:17,320
entry for a lot of people.

186
00:17:17,320 --> 00:17:20,360
Yeah and I would think that I mean it sounds like you've already done your homework in this

187
00:17:20,360 --> 00:17:25,840
area but I mean for the stuff that you want to do with your business right you are going

188
00:17:25,840 --> 00:17:32,000
to have something to where you can append updated data to a centralized location right

189
00:17:32,000 --> 00:17:39,120
on a regular basis I would guess and yeah I guess the key phrase there that you said

190
00:17:39,120 --> 00:17:41,720
is cost-effective.

191
00:17:41,720 --> 00:17:48,120
So with the OpenTHC project do they currently have all the data you were able to procure

192
00:17:48,120 --> 00:17:51,560
is that loaded up into their database?

193
00:17:51,560 --> 00:17:59,480
Yes essentially so actually David's the one who shared the data set with me so they have

194
00:17:59,480 --> 00:18:11,800
it there it's a work in development so it may not always be like easily accessible so

195
00:18:11,800 --> 00:18:18,880
to speak but David's done some work to you know make it so you can just get like a JSON

196
00:18:18,880 --> 00:18:25,160
dump I'm not 100% sure if I'll have to after this call I'll double check with David and

197
00:18:25,160 --> 00:18:27,720
actually share his contact information with you.

198
00:18:27,720 --> 00:18:29,520
Oh thank you.

199
00:18:29,520 --> 00:18:37,400
Because that's essentially what he's done is just put all this data on his server and

200
00:18:37,400 --> 00:18:40,520
he's trying to make it accessible through an API.

201
00:18:40,520 --> 00:18:41,520
Okay.

202
00:18:41,520 --> 00:18:47,120
But like I said yeah the current state of things but.

203
00:18:47,120 --> 00:18:54,600
Yeah I know I could like for these four tables that you walked me through I would you know

204
00:18:54,600 --> 00:18:59,360
in order to do the complete market basket analysis I would have to be able to either

205
00:18:59,360 --> 00:19:05,320
take all those tables and put them in a location through like you know Firebase or something

206
00:19:05,320 --> 00:19:11,800
I may have to end up paying for it I don't know but I'll need to get that yeah I'll need

207
00:19:11,800 --> 00:19:18,960
to get that data somewhere altogether that I could query subsets or just make general

208
00:19:18,960 --> 00:19:26,760
queries against but it sounds like you know if David is agreeable to it then maybe I could

209
00:19:26,760 --> 00:19:30,080
pull query data from his data set.

210
00:19:30,080 --> 00:19:39,800
Yes so I'll get you in touch with him you may have to kind of what's the phrase I don't

211
00:19:39,800 --> 00:19:46,000
know be the squeaky wheel or something but you know just to kind of you know you just

212
00:19:46,000 --> 00:19:52,640
to tell him like okay you know because he's a his whole philosophy is you know not to

213
00:19:52,640 --> 00:19:56,920
do a lot of development until there's like an actual need for it right so you don't want

214
00:19:56,920 --> 00:19:57,920
to be.

215
00:19:57,920 --> 00:20:03,440
Yeah oh yeah I wouldn't want to burden him with anything I again you know however I can

216
00:20:03,440 --> 00:20:08,400
get the data for this project this school project in one place I guess that's the main

217
00:20:08,400 --> 00:20:13,880
driver for me and if it was temporary I could set something up temporary and if you know

218
00:20:13,880 --> 00:20:18,360
but again you know it comes down to the cost effectiveness of it I definitely can't afford

219
00:20:18,360 --> 00:20:25,840
to stand the whole thing up for a bunch of people to bang against it but you know we'll

220
00:20:25,840 --> 00:20:26,840
see.

221
00:20:26,840 --> 00:20:31,480
Well yeah I'll get you in touch because like I said he's basically just serving it from

222
00:20:31,480 --> 00:20:42,720
like his box at his office and I think he's just got it wired up essentially through an

223
00:20:42,720 --> 00:20:43,720
API.

224
00:20:43,720 --> 00:20:47,200
Yeah under the table kind of thing.

225
00:20:47,200 --> 00:20:50,920
Well I think it's basically the type of thing where you basically just hit his URL with

226
00:20:50,920 --> 00:20:55,760
a dot JSON and you'll get a huge data dump.

227
00:20:55,760 --> 00:21:04,760
I'll get you in touch with him because that's how I got a hold of a lot of the data myself.

228
00:21:04,760 --> 00:21:06,720
Okay well thank you for that.

229
00:21:06,720 --> 00:21:10,760
Are you using Windows or Mac or Linux?

230
00:21:10,760 --> 00:21:11,760
Windows.

231
00:21:11,760 --> 00:21:24,520
Windows okay you know there's you could use WAMP which is Windows, Apache, MySQL, PHP

232
00:21:24,520 --> 00:21:30,240
but you could run like a local WAMP server on your PC even you can even run it on like

233
00:21:30,240 --> 00:21:39,600
a laptop and then have like an external drive and have MySQL database running and you know

234
00:21:39,600 --> 00:21:44,920
and you know just you know get some cheap external drive and put it on there.

235
00:21:44,920 --> 00:21:49,080
That's a good idea.

236
00:21:49,080 --> 00:21:53,680
Yeah so do one large download of the data put on it an external drive and then you're

237
00:21:53,680 --> 00:21:57,320
saying run WAMP against that external drive.

238
00:21:57,320 --> 00:22:02,320
Yeah or I mean you could probably even use you could probably even write you could probably

239
00:22:02,320 --> 00:22:06,040
put an access database on that external drive.

240
00:22:06,040 --> 00:22:09,720
Yeah unfortunately there's a two gig limit on access.

241
00:22:09,720 --> 00:22:10,720
Oh okay.

242
00:22:10,720 --> 00:22:14,640
Yeah yeah yeah that's one of the things.

243
00:22:14,640 --> 00:22:15,640
Yeah maybe WAMP.

244
00:22:15,640 --> 00:22:16,640
Okay.

245
00:22:16,640 --> 00:22:21,880
Yeah WAMP is free and then there's like a WAMP probe out you probably don't need that

246
00:22:21,880 --> 00:22:22,880
so.

247
00:22:22,880 --> 00:22:26,240
Okay I'll check that out that sounds like a good alternative too.

248
00:22:26,240 --> 00:22:27,240
Yeah.

249
00:22:27,240 --> 00:22:28,240
Yeah okay.

250
00:22:28,240 --> 00:22:37,240
No thanks for yeah thanks for all the advice on that.

251
00:22:37,240 --> 00:22:45,000
So I guess like what type of I'm just wondering like so what would like an example query be

252
00:22:45,000 --> 00:22:51,480
just so I can start thinking about like how you how you need to approach the data.

253
00:22:51,480 --> 00:22:56,400
Yeah I mean just this initial foray into the access database I mean really it's just a

254
00:22:56,400 --> 00:23:02,760
very simple straightforward selects you know certain fields from and then joining on the

255
00:23:02,760 --> 00:23:10,120
four tables on the primary keys so very straightforward simple query and that's really it.

256
00:23:10,120 --> 00:23:16,240
I would I think I'm definitely going to have to like I say break down those or create some

257
00:23:16,240 --> 00:23:23,360
subcategories of products and create a lookup table but really it's the query would be very

258
00:23:23,360 --> 00:23:28,200
straightforward very simple of course it's going to bring back a lot of data but yeah

259
00:23:28,200 --> 00:23:38,800
it's all very simple stuff.

260
00:23:38,800 --> 00:23:44,480
Question I guess with the market ballot basket analysis like what's like the like the measure

261
00:23:44,480 --> 00:23:48,880
so is it like the proportion of items?

262
00:23:48,880 --> 00:23:55,120
We were talking about that last time let me see if I can find something real quick to

263
00:23:55,120 --> 00:23:59,920
share with you guys online and maybe I can talk through it real quick.

264
00:23:59,920 --> 00:24:03,240
Well I gotta run I'll see you guys week.

265
00:24:03,240 --> 00:24:11,560
Charles definitely enjoy the IO conference tell me how what's new with tensorflow because

266
00:24:11,560 --> 00:24:15,840
I think there's some good uses for it that I'm missing out on so I'd love to get filled

267
00:24:15,840 --> 00:24:16,840
in by using it.

268
00:24:16,840 --> 00:24:19,840
Okay cool all right see ya.

269
00:24:19,840 --> 00:24:21,840
See ya Charles.

270
00:24:21,840 --> 00:24:29,920
Hey Keith I'm just going to look something up and maybe I can run you through some of

271
00:24:29,920 --> 00:24:38,600
the it's very actually very simple stuff and I'm but simple sometimes can be good so.

272
00:24:38,600 --> 00:24:44,840
That's actually analytics is like that's our one of our philosophies is you know simple

273
00:24:44,840 --> 00:24:47,800
is better than complex.

274
00:24:47,800 --> 00:24:54,760
Absolutely maybe this will help I might have to jump around on a bunch of stuff but let

275
00:24:54,760 --> 00:24:58,560
me share this with you.

276
00:24:58,560 --> 00:25:11,680
Where are we here where is the meetup?

277
00:25:11,680 --> 00:25:34,680
Sure this I'm going to bring this one over.

278
00:25:34,680 --> 00:25:35,680
Okay so.

279
00:25:35,680 --> 00:25:37,160
This you this looks fun.

280
00:25:37,160 --> 00:25:44,880
Oh yeah I was hang gliding.

281
00:25:44,880 --> 00:25:45,880
Yeah get my.

282
00:25:45,880 --> 00:25:49,000
Did you get a good thrill for that for sure.

283
00:25:49,000 --> 00:25:54,800
Yeah no I have an aviation background so I've always been interested in that kind of stuff.

284
00:25:54,800 --> 00:25:55,800
Cool.

285
00:25:55,800 --> 00:26:03,240
I just I just pulled this up is just so happened to come up here but um let's see what would

286
00:26:03,240 --> 00:26:08,800
be a good way to just start off with this here maybe.

287
00:26:08,800 --> 00:26:16,080
So with all the different combinations of products that you could have you have this

288
00:26:16,080 --> 00:26:22,400
is the left hand side or right hand side or what's called an antecedent and a consequent.

289
00:26:22,400 --> 00:26:28,560
And so when you have these antecedents and consequences you come up with these association

290
00:26:28,560 --> 00:26:29,560
rules.

291
00:26:29,560 --> 00:26:32,320
So here got ice cream and soda.

292
00:26:32,320 --> 00:26:37,440
You can also have a right hand side rule that has more than just soda in it.

293
00:26:37,440 --> 00:26:38,920
It could have several different things.

294
00:26:38,920 --> 00:26:42,560
So you say what's the relationship between ice cream and this group of products.

295
00:26:42,560 --> 00:26:47,220
Right so you get that relationship as well.

296
00:26:47,220 --> 00:26:52,720
But the most simplistic measure is what you're talking about it's called support and it's

297
00:26:52,720 --> 00:26:54,620
just proportionality.

298
00:26:54,620 --> 00:27:02,480
So how many how of all the combinations you can have through an entire product set what

299
00:27:02,480 --> 00:27:09,760
proportion of this rule this one at the top is it appears in the entire possible set of

300
00:27:09,760 --> 00:27:10,760
combinations.

301
00:27:10,760 --> 00:27:14,120
So here our support is point zero seven percent.

302
00:27:14,120 --> 00:27:15,120
Okay.

303
00:27:15,120 --> 00:27:21,920
And this is not unusual because there's so many different permutations of the association

304
00:27:21,920 --> 00:27:22,920
rules.

305
00:27:22,920 --> 00:27:26,640
So we get these small percentages but that's normal.

306
00:27:26,640 --> 00:27:29,600
So here you can just say that support is like popularity.

307
00:27:29,600 --> 00:27:36,120
So this ice cream and soda is a pretty popular combination.

308
00:27:36,120 --> 00:27:41,520
I'm going to go over here to lift and then I forgot the calculation maybe it's in here

309
00:27:41,520 --> 00:27:47,920
but there's another measurement called lift and lift is really you can think of as the

310
00:27:47,920 --> 00:27:52,080
strength of association between these two elements in the rule.

311
00:27:52,080 --> 00:27:56,480
It's the strength of association between ice cream and soda.

312
00:27:56,480 --> 00:28:01,760
So really if you look at think of this as a two dimensional plot where you have support

313
00:28:01,760 --> 00:28:09,600
on your y axis lift on your x axis what you want is a product that has high support in

314
00:28:09,600 --> 00:28:16,520
other words very popular and the association between those two items in your rule set is

315
00:28:16,520 --> 00:28:18,460
also very strong.

316
00:28:18,460 --> 00:28:23,160
So you're looking for in a x y axis you're looking for those data points in the upper

317
00:28:23,160 --> 00:28:25,280
right hand quadrant.

318
00:28:25,280 --> 00:28:29,680
So they're they're popular and there's a strong association.

319
00:28:29,680 --> 00:28:35,960
Those are the two main those are the two main metrics that I would be using.

320
00:28:35,960 --> 00:28:40,760
And I forgot what the confidence is for and there's actually there's go ahead.

321
00:28:40,760 --> 00:28:47,200
Can you see what lift was for again what's the capturing lift is like the strength of

322
00:28:47,200 --> 00:28:51,560
association between the items in the rule set.

323
00:28:51,560 --> 00:28:57,160
So ice cream and soda are strongly associated.

324
00:28:57,160 --> 00:29:03,560
So you want a rule that is high in support which is very popular but has high lift as

325
00:29:03,560 --> 00:29:04,560
well.

326
00:29:04,560 --> 00:29:07,260
The association of the rules are very strong.

327
00:29:07,260 --> 00:29:13,320
So you can count on that as a very popular strong rule and that's you know if you get

328
00:29:13,320 --> 00:29:17,320
several of these different rules together you can start thinking about product placement

329
00:29:17,320 --> 00:29:18,320
right.

330
00:29:18,320 --> 00:29:20,360
Ice cream and soda.

331
00:29:20,360 --> 00:29:27,920
So exactly so you could have high lift but low support so you could have someone just

332
00:29:27,920 --> 00:29:34,880
a couple people get topicals and capsules together every time but there's only like

333
00:29:34,880 --> 00:29:36,960
a small number of people that do that.

334
00:29:36,960 --> 00:29:40,840
So that would be high lift but low support.

335
00:29:40,840 --> 00:29:48,440
Yeah and what you find is you get a situation where in your X Y axis you in the upper right

336
00:29:48,440 --> 00:29:51,640
hand quadrant is where you want their association to be.

337
00:29:51,640 --> 00:29:57,080
But what typically happens is hardly anything is up in the right hand upper right hand quadrant

338
00:29:57,080 --> 00:30:04,160
is almost always empty and you get kind of this trade off between support and lift.

339
00:30:04,160 --> 00:30:08,640
And so there's a lot of human decision making that has to go into this.

340
00:30:08,640 --> 00:30:12,200
You know obviously you have to bring a lot of sales context into it right.

341
00:30:12,200 --> 00:30:18,440
Your subject matter expertise of the market and utilize this information based on let's

342
00:30:18,440 --> 00:30:19,600
say your circumstances.

343
00:30:19,600 --> 00:30:27,760
So let's say we shared this information with somebody you know it's running a store and

344
00:30:27,760 --> 00:30:32,680
you share their metrics for their particular store and you can say look there's a lot of

345
00:30:32,680 --> 00:30:35,260
customers that are buying these products together.

346
00:30:35,260 --> 00:30:41,840
Why not change your arrangement in the store to where you're placing them side by side.

347
00:30:41,840 --> 00:30:47,360
Or maybe you see a relationship that has very strong lift.

348
00:30:47,360 --> 00:30:51,840
They're highly associated with each other but they're not very popular.

349
00:30:51,840 --> 00:30:56,480
They don't have as much support but maybe if you take those two items and display them

350
00:30:56,480 --> 00:31:00,700
next to each other you may actually influence their popularity.

351
00:31:00,700 --> 00:31:05,260
So there's a lot of marketing kind of decisions that you could make out of this.

352
00:31:05,260 --> 00:31:08,760
You could change the arrangement of a store layout.

353
00:31:08,760 --> 00:31:14,360
You could actually have promotional coupons or whatever you know sales of the week or

354
00:31:14,360 --> 00:31:20,200
month or all kinds of different ideas that you could implement within a shopping setting.

355
00:31:20,200 --> 00:31:25,640
And that's that's and this has been around for a very long time and there's other variants

356
00:31:25,640 --> 00:31:27,000
of this approach.

357
00:31:27,000 --> 00:31:33,560
But I thought well gosh you know this is a new data set and there might be some opportunities.

358
00:31:33,560 --> 00:31:37,880
I mean if you said you were talking about producing white papers and things like this

359
00:31:37,880 --> 00:31:43,520
but from a analytics perspective maybe this is a service that you could offer to some

360
00:31:43,520 --> 00:31:52,560
of your customers or maybe you know maybe there's some I don't know maybe there's some

361
00:31:52,560 --> 00:31:58,040
sort of inventory type stuff you could do between the people that hold inventory and

362
00:31:58,040 --> 00:32:02,560
the things that they're going to try to sell to different retail outlets.

363
00:32:02,560 --> 00:32:07,440
I don't know there's lots of ways to think about this.

364
00:32:07,440 --> 00:32:12,440
This is brilliant Paul and like I loved your idea because it's essentially you're taking

365
00:32:12,440 --> 00:32:19,600
this you know this model this marketing technique analysis that's been around for a long time

366
00:32:19,600 --> 00:32:27,680
and like you said probably I mean maybe either no one or very few people has actually done

367
00:32:27,680 --> 00:32:29,720
this with Canvas.

368
00:32:29,720 --> 00:32:33,760
Yeah yeah it's like most things in life right.

369
00:32:33,760 --> 00:32:38,640
I mean here you are you know tell me about your story and your journey right which is

370
00:32:38,640 --> 00:32:39,640
really cool.

371
00:32:39,640 --> 00:32:46,040
I mean you've got a developer's background but you found yourself in a completely new

372
00:32:46,040 --> 00:32:54,000
industry in some of the lab work right and so it's that combination of interesting intersection

373
00:32:54,000 --> 00:32:56,360
points which provides opportunity.

374
00:32:56,360 --> 00:33:02,040
I don't think I'm not a big believer in swinging for the fences and home runs.

375
00:33:02,040 --> 00:33:07,200
I think those are freaks of nature for the most part but if you position yourself right

376
00:33:07,200 --> 00:33:12,800
in the right kind of circumstances you can develop opportunities just by connecting disparate

377
00:33:12,800 --> 00:33:18,880
dots and it's kind of like where you are right so and I thought well gosh you know this is

378
00:33:18,880 --> 00:33:27,240
a again very standard old technique but hey let's apply it in a different environment.

379
00:33:27,240 --> 00:33:31,040
It's a good idea and I like your idea about putting it out there for people to use so

380
00:33:31,040 --> 00:33:39,280
what came to mind was essentially almost like a like an online calculator where I don't

381
00:33:39,280 --> 00:33:43,800
know maybe you know the company could put in their license or what they're trying to

382
00:33:43,800 --> 00:33:50,320
look at but and then they could do this market basket analysis for that licensee.

383
00:33:50,320 --> 00:33:54,560
Exactly yeah they could do for the licensee.

384
00:33:54,560 --> 00:34:00,520
They could have this online calculator that that may make suggestions for cross promotional

385
00:34:00,520 --> 00:34:03,520
products.

386
00:34:03,520 --> 00:34:08,560
There could be all sorts of different services that you could offer electronically and sure

387
00:34:08,560 --> 00:34:15,120
is better to develop things and dealing with electronic products like this as opposed to

388
00:34:15,120 --> 00:34:16,360
tangible ones right.

389
00:34:16,360 --> 00:34:20,880
There's so much less oh there's a lot less overhead in developing a business that way

390
00:34:20,880 --> 00:34:26,080
if you just deal with the information so yeah that could be some services this could act

391
00:34:26,080 --> 00:34:37,600
as a service to for folks out there right cross promotional couponing or whatever but

392
00:34:37,600 --> 00:34:41,400
yeah so that's that's the intent of this thing again it's just a matter of me being able

393
00:34:41,400 --> 00:34:48,640
to get all the data that I need that's accessible that's going to probably be the most work

394
00:34:48,640 --> 00:34:54,640
I would think because after I've got that the actual algorithm is pretty simple to run

395
00:34:54,640 --> 00:34:58,120
and then we can generate a lot of different kinds of charts and that's when it starts

396
00:34:58,120 --> 00:35:01,000
getting fun when you can just explore right.

397
00:35:01,000 --> 00:35:05,920
Like if let's say you wanted to do a regional market basket analysis right maybe there's

398
00:35:05,920 --> 00:35:12,040
certain parts of the state that I don't know they might have certain products that caters

399
00:35:12,040 --> 00:35:15,240
their demographic I don't know.

400
00:35:15,240 --> 00:35:21,800
I think that could that could be interesting because that is actually something I just

401
00:35:21,800 --> 00:35:27,080
heard word of mouth was like I think the company was talking about like in Illinois or something

402
00:35:27,080 --> 00:35:33,640
how in different towns there they actually sell different products so.

403
00:35:33,640 --> 00:35:38,920
Okay yeah that seems to make sense right I mean you definitely have regional products

404
00:35:38,920 --> 00:35:49,720
in anything yeah so but you know it's great that you're doing this because having having

405
00:35:49,720 --> 00:35:56,400
as you well know having the data centralized in an accessible way for people to get to

406
00:35:56,400 --> 00:36:03,240
is is key to everything right and if you can in a systematic way you know it sounds like

407
00:36:03,240 --> 00:36:08,280
that's what kind of the open THC project is trying to do too but to be able to have that

408
00:36:08,280 --> 00:36:15,400
all together one spot puts you like right in the hub of everything you know and as a

409
00:36:15,400 --> 00:36:21,600
facilitator and as an enabler you know to empower people to do this type of work I mean

410
00:36:21,600 --> 00:36:24,400
it's a great place to be.

411
00:36:24,400 --> 00:36:30,640
Because exactly because that's that is it was originally for labs but really anyone

412
00:36:30,640 --> 00:36:36,880
can do this industry that's analytics is philosophy is that if that it can be accessible it's

413
00:36:36,880 --> 00:36:42,640
just going to make everyone's lives better and you know business is going to run better

414
00:36:42,640 --> 00:36:52,160
absolutely like I mean technically the data is there but I mean as you've found out and

415
00:36:52,160 --> 00:37:03,200
as I know it's not easy to work with you've got these giant TSVs that are hard to parse

416
00:37:03,200 --> 00:37:11,240
and they're too big to open and so yeah I think this is going to be a good project to

417
00:37:11,240 --> 00:37:20,160
work on and I think there's going to be some obviously you're you know the last week I've

418
00:37:20,160 --> 00:37:25,800
had to look at the data the little bit I've seen to your point it is yeah you just have

419
00:37:25,800 --> 00:37:30,980
to know what's going on in there to use it the only way to do that is to just to suffer

420
00:37:30,980 --> 00:37:38,480
the slings and arrows or map it out you know and then and share that with people but yeah

421
00:37:38,480 --> 00:37:45,800
no it's such an interesting space that what's so exciting about it is and kudos to you is

422
00:37:45,800 --> 00:37:52,520
that to get in there you know as early as you can and be some of the first people in

423
00:37:52,520 --> 00:37:57,840
there because that's where the low-hanging fruit is and this is an example of low-hanging

424
00:37:57,840 --> 00:38:02,880
fruit in my opinion right it's something that we can do pretty easily after we get to the

425
00:38:02,880 --> 00:38:07,440
data it's pretty easily and start playing around results and once we get this set up

426
00:38:07,440 --> 00:38:12,160
you could do the same thing and I'm going to be using R because I'm just more proficient

427
00:38:12,160 --> 00:38:18,640
in R but you know for your perspective you could use you know Python or what's it called

428
00:38:18,640 --> 00:38:23,680
what's the data science package is it SciPy or I can't remember what it's called but you

429
00:38:23,680 --> 00:38:32,640
can do you could yeah there's a couple stats models but it yeah I'm a big fan of it's you

430
00:38:32,640 --> 00:38:38,440
know your your weapon of choice so yeah exactly yeah so I mean you could do that you could

431
00:38:38,440 --> 00:38:43,280
do the equivalent thing of what I'm trying to do if just by you know using some simple

432
00:38:43,280 --> 00:38:50,000
Python models and get the same results and play around with it make maybe make those

433
00:38:50,000 --> 00:38:54,720
results like you're saying you know or make it accessible through an API where people

434
00:38:54,720 --> 00:38:59,720
can just run some inputs and get some outputs that might be the beginning of it you know

435
00:38:59,720 --> 00:39:04,040
this is an exploratory experience right just just trying to figure out how to get to the

436
00:39:04,040 --> 00:39:07,840
data and what we can pull out of the data and whatever you learn from it you know that

437
00:39:07,840 --> 00:39:15,400
kind of kind of inform you on directions you want to take as well I think I think it's

438
00:39:15,400 --> 00:39:23,800
worth pursuing because the data is there everybody wants it and it's just just slightly out of

439
00:39:23,800 --> 00:39:30,720
reach of everybody and then this is the amount of work you have to put into what you've already

440
00:39:30,720 --> 00:39:38,200
done right is you know through your API calls and through the the dictionary object that

441
00:39:38,200 --> 00:39:44,280
you sent me where you see those relationships that's the key those relationships of how

442
00:39:44,280 --> 00:39:49,360
that everything's tied together that's the key and it takes a lot of work to untangle

443
00:39:49,360 --> 00:39:56,240
that and create that comprehensive picture and that's that's the barrier to entry well

444
00:39:56,240 --> 00:40:05,640
I can help there because it just took a lot of banging my head against the wall so this

445
00:40:05,640 --> 00:40:09,920
was you know I purchased from the lab results and so we had all these lab results that needed

446
00:40:09,920 --> 00:40:15,120
to be posted and then like you know unhappy clients that are saying well where are our

447
00:40:15,120 --> 00:40:26,160
lab results so right but it's it's just a lot there but like I said there's a

448
00:40:26,160 --> 00:40:30,760
couple like key fields that kind of connect everything together so you know if you know

449
00:40:30,760 --> 00:40:39,560
what you're looking for it's not that overwhelming so yeah I think it's worth putting together

450
00:40:39,560 --> 00:40:47,240
because like you said I mean just retailers alone would just have so much value plus you've

451
00:40:47,240 --> 00:40:58,120
got cultivators processors the labs independent researchers you know the government body yeah

452
00:40:58,120 --> 00:41:05,240
and there's one one one sorry to interrupt you but there's my last class which was on

453
00:41:05,240 --> 00:41:13,840
optimization there's models that you can develop for transportation right like optimizing optimizing

454
00:41:13,840 --> 00:41:21,760
routes right so if you went to a company and they if you could find out I haven't seen

455
00:41:21,760 --> 00:41:28,980
the transportation data but if it has point a to point b routes in it you can create optimization

456
00:41:28,980 --> 00:41:36,080
patterns for these companies where they're spending less time on the road in fact that's

457
00:41:36,080 --> 00:41:41,880
one of the most interesting that's one of the areas that I find the most interesting

458
00:41:41,880 --> 00:41:48,680
so when I worked at the laboratory maybe a quarter of it was posting the lab results

459
00:41:48,680 --> 00:41:55,320
three quarters of it was dealing with inventory transfers so that's primarily we were receiving

460
00:41:55,320 --> 00:42:04,220
them and so like you said it's a scheduling problem so scheduling problem yes so at labs

461
00:42:04,220 --> 00:42:16,280
they send out couriers so you'd send out two or three couriers they do I don't know three

462
00:42:16,280 --> 00:42:23,840
to three to ten stops a piece you know like you said you can you can model all of that

463
00:42:23,840 --> 00:42:31,000
so you can do there's the the traveling salesman problem with one driver and then the vehicle

464
00:42:31,000 --> 00:42:37,520
routing problem which is multiple drivers and I think that's actually like a classic

465
00:42:37,520 --> 00:42:43,520
optimization problem because I don't think it's like it's obviously not close-ended and

466
00:42:43,520 --> 00:42:49,660
you often have to just do a brute force approach where you just calculate the distance between

467
00:42:49,660 --> 00:42:57,580
all the routes yeah yeah some of the things that we're using and we cover this stuff so

468
00:42:57,580 --> 00:43:01,840
fast in class that I just barely got through it with the skin of my teeth but we were using

469
00:43:01,840 --> 00:43:09,840
like genetic algorithms to kind of hone in on on some of the most likely answers and

470
00:43:09,840 --> 00:43:14,540
we did we actually went over the traveling salesman problem several times you know so

471
00:43:14,540 --> 00:43:21,980
yeah and maybe you know maybe that's not such a great example because you can buy off-the-shelf

472
00:43:21,980 --> 00:43:28,380
software for scheduling but again that you could you could still offer that as a service

473
00:43:28,380 --> 00:43:33,280
and based on your experience it sounds like it's a legit problem but you have to remind

474
00:43:33,280 --> 00:43:39,240
yourself like I mean you don't have access to Amazon's data you know you don't have access

475
00:43:39,240 --> 00:43:46,020
to the UPS's data where they're zipping all over the place but here we actually have time

476
00:43:46,020 --> 00:43:54,740
stamped transfers they have the route so you have point A point B we left at time A well

477
00:43:54,740 --> 00:44:05,420
it's estimated we left at time A estimated arrival at time B so I mean there's just interesting

478
00:44:05,420 --> 00:44:09,980
like you said there's low-hanging fruit so you can just do summary statistics like how

479
00:44:09,980 --> 00:44:20,340
many miles were driven yeah like like that would be just an interesting data point so

480
00:44:20,340 --> 00:44:26,740
like like yeah like how many and then you can just estimate like you know how much gas

481
00:44:26,740 --> 00:44:33,580
they're spending or who knows but there are transportation companies and so I don't know

482
00:44:33,580 --> 00:44:39,780
the particular questions to ask off the top of my head but I think it's a yet another

483
00:44:39,780 --> 00:44:49,900
case where you've got a unique data set that you just don't find in other industries and

484
00:44:49,900 --> 00:44:59,580
yeah I mean to your point yeah I mean to your point with you know the travel salesman stuff

485
00:44:59,580 --> 00:45:03,600
and the miles driven summary statistics I'm sure there's probably insurance companies

486
00:45:03,600 --> 00:45:08,040
would like to know that who are insuring these these transporters right they mean they may

487
00:45:08,040 --> 00:45:16,260
want to know more have more detailed information about that there's so many ways that I mean

488
00:45:16,260 --> 00:45:22,860
you know I may be overly optimistic I'm an optimistic person by nature but I mean just

489
00:45:22,860 --> 00:45:28,220
just be ready because it's looking from the outside view and for the short time that I've

490
00:45:28,220 --> 00:45:33,660
known you it seems like there's going to be a lot of opportunities coming your way and

491
00:45:33,660 --> 00:45:37,020
you're probably going to have to like figure out how to spend your time on the most important

492
00:45:37,020 --> 00:45:43,060
ones but yeah no it just there's tons of opportunity here and I know there's other people that

493
00:45:43,060 --> 00:45:48,700
are jumping into this well other people that are companies I mean I've you know it was

494
00:45:48,700 --> 00:45:53,220
the name that analytics company there's several of them but there's one that stood out to

495
00:45:53,220 --> 00:46:04,140
me and if I heard their name I remember it but the two big ones are like new frontier

496
00:46:04,140 --> 00:46:19,140
data and I kept on seeing Google I kept on seeing Google searches but I remember it but

497
00:46:19,140 --> 00:46:24,380
I mean obviously there's people that are you know ahead and in this space but you know

498
00:46:24,380 --> 00:46:28,900
the the devil's in the details and you're really getting deep into the Washington State

499
00:46:28,900 --> 00:46:34,560
data you're familiar with it I can see a time in the future where there's probably going

500
00:46:34,560 --> 00:46:39,380
to be some big money players that are going to want to consolidate all the different states

501
00:46:39,380 --> 00:46:45,100
so they can get that you know 360 view of what's going on and that's where the I mean

502
00:46:45,100 --> 00:46:50,940
that's where the the big money I would think would be a big opportunity but I could see

503
00:46:50,940 --> 00:46:55,900
that happen in one day where you know if you're doing pretty well with analytics I could see

504
00:46:55,900 --> 00:47:00,500
somebody that's got like well-financed that might knock on your door and go hey how much

505
00:47:00,500 --> 00:47:05,380
do you want to sell analytics for you know I could see that happening it doesn't fit

506
00:47:05,380 --> 00:47:11,020
in any industry automotive was like this at the turn of the last century and and any you

507
00:47:11,020 --> 00:47:15,300
know IT different companies you start off with like tons of little companies and then

508
00:47:15,300 --> 00:47:22,700
there's the consolidation phase so yeah I could see that happening and I think you're

509
00:47:22,700 --> 00:47:30,340
right so essentially what you've got now is a shortage of data analytics so everyone needs

510
00:47:30,340 --> 00:47:37,420
it there's a couple players providing it but they're they can't provide enough they're

511
00:47:37,420 --> 00:47:42,300
trying they're trying their best and they are producing a lot but like you said there's

512
00:47:42,300 --> 00:47:49,540
so many avenues like if you've got somebody looking at I mean they're looking at like

513
00:47:49,540 --> 00:47:55,340
demographics but you know you're looking at market back market basket analysis that's

514
00:47:55,340 --> 00:48:03,580
those are two different analyses then if you toss in you know logistics that's a whole

515
00:48:03,580 --> 00:48:11,260
other ballpark yeah and that's these are all huge areas right they're all huge segments

516
00:48:11,260 --> 00:48:17,900
of this whole process there's what's great about the seat to sale process is it's just

517
00:48:17,900 --> 00:48:26,140
an immense supply chain really supply chain right and there's just so much opportunity

518
00:48:26,140 --> 00:48:34,020
in that space so I was going to ask you based on your your history in the labs have you

519
00:48:34,020 --> 00:48:42,060
ever come across people that propagate from it's basically from a cell cutting from a

520
00:48:42,060 --> 00:48:48,380
plant and they grow them in like kind of like is it Algar and like at least Petri dishes

521
00:48:48,380 --> 00:48:57,940
it's like string good do you mean like like clones so yeah things of the plant well not

522
00:48:57,940 --> 00:49:03,780
a snipping of the plant and then you just prop you know propagate it in some roots nutrients

523
00:49:03,780 --> 00:49:10,340
but I've heard that some people can actually take cell samples from the plants from like

524
00:49:10,340 --> 00:49:16,180
some place certain place on the stem put them in a Petri dish with this Algar kind of growing

525
00:49:16,180 --> 00:49:22,100
medium and then the cells will grow and also a plant will start to grow and the idea is

526
00:49:22,100 --> 00:49:28,980
that any kind of viruses and other types of things are left behind in this process because

527
00:49:28,980 --> 00:49:34,600
it's done in a sterile environment so you can maintain your strain without bringing

528
00:49:34,600 --> 00:49:39,180
all the baggage along I guess that's kind of I don't know if you've heard of that I

529
00:49:39,180 --> 00:49:44,740
just I stumbled on to and I was like huh sounds interesting that's that's an incredibly interesting

530
00:49:44,740 --> 00:49:50,580
technique I haven't heard of that but like you said I think I wouldn't be surprised if

531
00:49:50,580 --> 00:49:55,820
people are doing that because people are getting quite innovative they're also a little secretive

532
00:49:55,820 --> 00:50:05,100
about those things so I think if someone yeah was using like that technique they may not

533
00:50:05,100 --> 00:50:14,420
be telling everybody I at the lab we primarily had people just asking questions about like

534
00:50:14,420 --> 00:50:20,580
you know different extraction techniques so that's well a lot of like sort of like the

535
00:50:20,580 --> 00:50:29,500
scientists who were doing the chemists they're looking for the the most efficient cost-efficient

536
00:50:29,500 --> 00:50:37,940
and highest yield extraction techniques okay that's the processors and I actually haven't

537
00:50:37,940 --> 00:50:41,740
heard too much about the cultivators like I said I think it may be because they're kind

538
00:50:41,740 --> 00:50:50,780
of guarding these secrets because from what I've heard it's like growing knowledge is

539
00:50:50,780 --> 00:50:57,060
actually kind of rare so you have a lot of people kind of seeking like a like a head

540
00:50:57,060 --> 00:51:01,900
grower like a head cultivator and like I guess everybody you know claims that they're like

541
00:51:01,900 --> 00:51:08,540
an expert sure it's really it's really tough to you know find people that actually know

542
00:51:08,540 --> 00:51:16,700
what they're doing so yeah yeah and I know that here in Michigan there's such a large

543
00:51:16,700 --> 00:51:22,300
black market that's starting to now come into the sunlight you get a lot of more legit businesses

544
00:51:22,300 --> 00:51:30,260
that are starting to bubble up and it's right now in the last 18 months it's been crazy

545
00:51:30,260 --> 00:51:34,860
just talking with my brother-in-law who's involved in the business what little I know

546
00:51:34,860 --> 00:51:40,500
about it I've heard through him and you get so many facilities that are popping up like

547
00:51:40,500 --> 00:51:47,820
mushrooms now and some of these people are pretty well financed to get into this but

548
00:51:47,820 --> 00:51:53,580
the businesses so let me give you a little bit of background on southeast Michigan the

549
00:51:53,580 --> 00:51:57,260
most the largest community that's getting involved with this they're called it's called

550
00:51:57,260 --> 00:52:02,220
the Chaldean community so there's a large middle eastern population here in southeast

551
00:52:02,220 --> 00:52:07,860
Michigan like for example my wife is Jordanian right so there's a lot of a lot of people

552
00:52:07,860 --> 00:52:13,660
from the middle east here they came to the auto to work in automotive industry and what's

553
00:52:13,660 --> 00:52:20,460
happened is following generations they're more of like a like a merchant class of people

554
00:52:20,460 --> 00:52:25,580
where they do they have a lot of restaurants gas stations this sort of thing so they've

555
00:52:25,580 --> 00:52:30,020
a lot of these folks are well financed have their own businesses and saw this as an opportunity

556
00:52:30,020 --> 00:52:36,020
to jump in so the Chaldean community owns the vast majority of the operations in southeast

557
00:52:36,020 --> 00:52:43,020
Michigan and but here's the thing they don't have the level of sophistication of the things

558
00:52:43,020 --> 00:52:47,540
that you and I are talking about they would have no idea about this type of thing and

559
00:52:47,540 --> 00:52:52,540
as these businesses become more and more mainstream and legit businesses which that's where it's

560
00:52:52,540 --> 00:52:56,640
all going there's going to be for the things that you're talking about like making things

561
00:52:56,640 --> 00:53:01,500
easier for people with compliance there's going to be a lot of opportunity I think and

562
00:53:01,500 --> 00:53:05,820
you probably already you know you've already experienced it with based on what you've told

563
00:53:05,820 --> 00:53:12,040
me so that's happening here it's going to happen everywhere right so it's just a matter

564
00:53:12,040 --> 00:53:20,140
of again low hanging fruit which things seem to be the most important exactly because like

565
00:53:20,140 --> 00:53:25,620
you said there's you know there's entrepreneurs business owners they're trying to set up shop

566
00:53:25,620 --> 00:53:34,220
and yeah they need people that they they have a bit of experience and they have some knowledge

567
00:53:34,220 --> 00:53:43,660
to share just you just your knowledge on your lab expertise I bet you could probably hang

568
00:53:43,660 --> 00:53:48,900
a shingle out and make your and people in southeast Michigan would want to consult with

569
00:53:48,900 --> 00:53:56,900
you I guarantee it go ahead well what would they want to consult about do you think I

570
00:53:56,900 --> 00:54:02,060
think the most basic question how do I pass the tests right what things should I be looking

571
00:54:02,060 --> 00:54:08,420
for yeah how do I set my operations where I can get my certifications done I mean that's

572
00:54:08,420 --> 00:54:15,220
the most basic basic question right so and you could advise them on well I have this

573
00:54:15,220 --> 00:54:18,780
five step plan or whatever it is right based on your experience these are the things that

574
00:54:18,780 --> 00:54:25,420
people mostly do not pass on and the reasons why that you could for an hour of consultation

575
00:54:25,420 --> 00:54:29,420
you could charge a thousand dollars or something right it's not it wouldn't be unusual to charge

576
00:54:29,420 --> 00:54:35,780
those types of fees I don't think because these people a lot of these people have the

577
00:54:35,780 --> 00:54:44,940
money to do that so you know just think about it just think about it I mean it might be

578
00:54:44,940 --> 00:54:49,160
an extra another revenue stream for you or something and if I could help you I'll let

579
00:54:49,160 --> 00:54:57,020
you know I could I could talk to my brother-in-law's and put it out there that there's a lab expert

580
00:54:57,020 --> 00:55:02,520
that I know and if if anybody you know if he's looking for consultation for specific

581
00:55:02,520 --> 00:55:09,500
questions he's a good person to talk to definitely I could definitely help out so there's always

582
00:55:09,500 --> 00:55:16,820
some of the things I know about I've seen I've seen my fair share of failures I I know

583
00:55:16,820 --> 00:55:22,580
I know the pain points of the producers and processors and generally can help just make

584
00:55:22,580 --> 00:55:29,700
things go a little easier and then the sugar on top is it also back it up with data and

585
00:55:29,700 --> 00:55:34,420
you can actually look at say Washington data and actually show them okay you know these

586
00:55:34,420 --> 00:55:39,180
are the failure rates in Washington maybe these are the failure rates in California

587
00:55:39,180 --> 00:55:44,240
you know this is this is about what you should you can expect right and you could say these

588
00:55:44,240 --> 00:55:49,780
are the failure rates and these are the reasons why they did fail and this is what you could

589
00:55:49,780 --> 00:55:57,260
do to mitigate those risks really yeah total consulting gig yeah that's brilliant Paul

590
00:55:57,260 --> 00:56:06,100
you're got a lot of ideas well I mean you just yeah you're in the right spot man I mean

591
00:56:06,100 --> 00:56:13,580
it's a good good spot to be in I think but yeah I mean I'll talk to my brother-in-law

592
00:56:13,580 --> 00:56:17,940
and feel a feel him out if you know what the demand would be for something like this and

593
00:56:17,940 --> 00:56:22,020
then next time we'll talk I'll just kind of hit you up or I'll just send you a note or

594
00:56:22,020 --> 00:56:28,740
whatever to find out if it seems like it would be a viable thing and then you could say well

595
00:56:28,740 --> 00:56:36,620
if it's something I want to do what kind of evidence presentations you know other kinds

596
00:56:36,620 --> 00:56:44,260
of things that I could use as consulting information and maybe you could do a test run you could

597
00:56:44,260 --> 00:56:48,980
do test run with my brother-in-law or something and say if you're trying to get yeah I mean

598
00:56:48,980 --> 00:56:54,420
then it's a no harm no foul and you can fumble your way through it and totally bomb it it

599
00:56:54,420 --> 00:57:00,820
won't matter you're just learning so maybe okay that could be a good experience because

600
00:57:00,820 --> 00:57:08,180
I could hear what you know what questions he has to ask and yeah do my best to answer

601
00:57:08,180 --> 00:57:13,500
yeah exactly so maybe that's something we could do just to yeah we should we should

602
00:57:13,500 --> 00:57:19,140
do that even if it was a totally informal and I got you guys together just to chat for

603
00:57:19,140 --> 00:57:25,700
half an hour you would learn about the southeast michigan market and he would learn from you

604
00:57:25,700 --> 00:57:34,060
on things that he could do from so he's got a what do you call it a caregiver grow operation

605
00:57:34,060 --> 00:57:44,620
is what he's got so and I know he wants to grow and literally grow his operation and

606
00:57:44,620 --> 00:57:48,260
get some licensing but in order to do that he's got to learn about how to get you know

607
00:57:48,260 --> 00:57:53,660
passed from the state so it could be a mutually beneficial thing so that's something to think

608
00:57:53,660 --> 00:57:59,380
about definitely I'm all aboard because that's what I'm here to do is help and share my knowledge

609
00:57:59,380 --> 00:58:04,260
with other people in the industry and so that they can yeah they can run their businesses

610
00:58:04,260 --> 00:58:10,780
better so I'm absolutely and don't don't forget you know if you know if this starts if these

611
00:58:10,780 --> 00:58:18,020
different areas start to become promising for you make sure you get paid for it definitely

612
00:58:18,020 --> 00:58:24,580
that's the first thing yeah exactly so but anyway all right I'm gonna have to probably

613
00:58:24,580 --> 00:58:33,180
let's bolt here yes I've got a full day ahead so thank you Paul yeah awesome ideas yeah

614
00:58:33,180 --> 00:58:39,100
it's always awesome talking with you and I'll stay in touch because I think what you've

615
00:58:39,100 --> 00:58:45,140
lined out in Michigan is promising and we've got a lot of promising work ahead with this

616
00:58:45,140 --> 00:58:51,140
data set yeah it sounds right it sounds really fun I'm looking forward to it and good luck

617
00:58:51,140 --> 00:58:57,380
with your move today definitely Paul thank you all right have an awesome day I'll talk

618
00:58:57,380 --> 00:59:24,460
to you later Keegan take care bye bye