These are chat archives for FreeCodeCamp/DataScience

28th
Oct 2018
Honman Yau
@honmanyau
Oct 28 2018 01:43
@bigyankarki No worries! I haven’t worked with the Python version of Tensorflow much (that probably would make some cringe) but it looks like you’ve found the issue. Good luck with the rest of it! :)
Alice Jiang
@becausealice2
Oct 28 2018 17:40
@pdurbin That's my bad. Thanks lol
@KiranGkiran11_twitter The answer depends on what your data looks like and what tools you're using
Alice Jiang
@becausealice2
Oct 28 2018 18:13
Another R question, if I have a dataframe like:
A 1
B c(2, 3, 4, 5)
C c(6, 7, 8)
D 9
(that being two columns, and my actual data is all strings, not integers)
How can I flatten the second column into a single vector? c(1,2,3,4,5,6,7,8,9)
@GoldbergData and @erictleung I know you're both the resident R gurus, I throw myself at your mercy
mstellaluna
@mstellaluna
Oct 28 2018 18:30
For anyone interested there is a Data Science/Machine Learning/Big Data/Analytics conference in November in Atlanta https://datascicon.tech/
i have no idea why i get these tweets but figured someone might be interested in here
Alice Jiang
@becausealice2
Oct 28 2018 18:32
You get these tweets for the purpose of sharing here :blush:
Thanks @mstellaluna it looks worth checking out for sure
mstellaluna
@mstellaluna
Oct 28 2018 18:32
i have no idea i think someone im following is following someone in DS cuz i shouldn't be getting them cuz i follow mostly Java people
Alice Jiang
@becausealice2
Oct 28 2018 18:33
Ah yeha I was getting mystery tweets that way for a while and when one of them crossed a line I went in and discovered there's a setting to turn it off
mstellaluna
@mstellaluna
Oct 28 2018 18:34
oh nice
i had one tweet that crossed the line, i just unfollowed the person that re-tweeted it.. was the end of that
Alice Jiang
@becausealice2
Oct 28 2018 18:36
Yeah, in my case they hadn't liked or retweeted, it was just because they follow them. I can't find the setting now, but I remember there's a bunch of timeline settings that are opt-out like seeing tweets by follows of follows and stuff like that
mstellaluna
@mstellaluna
Oct 28 2018 18:36
i have to look into that.. i probably have to login to the browser instead of mobile to see the settings.. i mostly use mobile versoin
Alice Jiang
@becausealice2
Oct 28 2018 18:38
OH! I think it's under "privacy and safety" and there's a setting inside the "Privacy" section for personalization that's all about finding content on your behalf
mstellaluna
@mstellaluna
Oct 28 2018 18:39
thanks ! i'll take a look later on tonight
but then i can't come in here spamming conferences i don't get commission out of lol
Alice Jiang
@becausealice2
Oct 28 2018 18:40
That's not the one I Was thinking of anyways. Just have a look through all Twitter settings. I was genuinely shocked at the type of settings the have that are opt-out
mstellaluna
@mstellaluna
Oct 28 2018 18:42
i'll hunt around the settings.. thanks again
Alice Jiang
@becausealice2
Oct 28 2018 18:42
:+1:
I'm all about that "trust no one" life
mstellaluna
@mstellaluna
Oct 28 2018 18:42
yup, same i get twitter followers who accounts have nothing.. immediate block
Alice Jiang
@becausealice2
Oct 28 2018 18:44
Even companies, though. I have a friend who was outed to his entire family because a notification popped up on his facebook that he had been tagged in someone else's pictures by the facebook algos of the pride parade.
It was like the Target pregnant teenager story, but felt a lot more personal and real lol
Alice Jiang
@becausealice2
Oct 28 2018 18:48
Oh my god I need that book in my life RIGHTNOW
There's a lot of examples of it, though
One of my favorites is Ben Taylor's attractiveness algorithm
pdurbin @pdurbin looks it up
Alice Jiang
@becausealice2
Oct 28 2018 19:13
@pdurbin Ben Taylor was working for HireVue, a company that does video interviews of job applicants to reduce human bias and increase diversity and experience in the workplace when he made an algorithm that determines how attractive someone is
sounds super sketchy, but they were able to use his algorithm to discover that because of the attractiveness bias in people, their algorithms also had an attractiveness bias (garbage in, garbage out)
Philip Durbin
@pdurbin
Oct 28 2018 19:28
Interesting stuff. I work on a data sharing platform so I think he should put his CSV in a repository rather than sticking it on Google Docs. At least it's a bitly link he can update. :)
Philip Durbin
@pdurbin
Oct 28 2018 19:49
Of course, it isn't his dataset. At least he links to where he got it.
Eric Leung
@erictleung
Oct 28 2018 19:57
Another good book on algorithmic and mathematical modeling bias is "Weapons of Math Destruction"
@becausealice2 I'm away from my computer, so I can't be sure it'll work, but you might be able to just pull the column itself and then do a reduce on making a vector. That might be a bit messy, but that's what's on the top of my head.
Philip Durbin
@pdurbin
Oct 28 2018 20:04
I tried a bit to get a dataframe to look like that but I don't R very well.
Alice Jiang
@becausealice2
Oct 28 2018 20:18
@pdurbin i don't either. At this point about 50% of my work on this data has been responses from me asking here and Twitter for help
and the other 50% has little to do with R code and is mostly a result of my supreme talent of usually knowing what to type into a google search
Philip Durbin
@pdurbin
Oct 28 2018 20:20
Is the data public?
Alice Jiang
@becausealice2
Oct 28 2018 20:23
It's not :(
I had to get clearance from US government to even look at it lol
(lowest level of clearances, I don't matter to the country, I just needed government approval because it's data from veterans)
Philip Durbin
@pdurbin
Oct 28 2018 20:30
ah
Alice Jiang
@becausealice2
Oct 28 2018 20:53
@erictleung That's done it, thanks! :D
Eric Leung
@erictleung
Oct 28 2018 21:34
@becausealice2 nice! :smile:
Data manipulation problems are interesting. I was really proud of this code I wrote in response to a question on Twitter https://twitter.com/ConnorJPSmith/status/1049735410300014592?s=09 I almost didn't think it was possible.
Alice Jiang
@becausealice2
Oct 28 2018 21:49
I almost don't think most of what I do is possible. That's why I call it "datamancy"