These are chat archives for FreeCodeCamp/DataScience

18th
May 2017
satenndrra
@satenndrra
May 18 2017 09:28
@apottr thanks for sharing, didn't know about this one! The wiki https://www.reddit.com/r/datasets/wiki/index actually list out the public datasets available.
CamperBot
@camperbot
May 18 2017 09:28
satenndrra sends brownie points to @apottr :sparkles: :thumbsup: :sparkles:
:star2: 2864 | @apottr |http://www.freecodecamp.com/apottr
satenndrra
@satenndrra
May 18 2017 09:32
I don't want to go too deep into data science (god, it's so vast!), but I just can't stop thinking about stuff like visualization and machine learning :smile:
satenndrra
@satenndrra
May 18 2017 09:41
Anyone here interested in music data visualization? Stuff like genre/ history visualization http://www.musicmap.info/ , http://svds.com/rockandroll/#thebeatles , https://research.google.com/bigpicture/music/
Hèlen Grives
@mesmoiron
May 18 2017 12:26
Well I ended up doing computational math; just to let the concepts sink in. Eventually I think it will come in handy when making your own algo's . But opions differ. Not everybody forges its own tools. Spreadsheets are widely used but mostly not for complicated stuff by many people.
evaristoc
@evaristoc
May 18 2017 13:43
@satenndrra @youlz @NDuma I am not sure if @PunkDado is available. Is any of you in for a project involving the 2017 Survey? Then we can work the visualizations together with @becausealice2 ?
Leonardo Raduy Lemos
@PunkDado
May 18 2017 13:53
@evaristoc I'm available, just maybe not so fast... ;-) Yesterday I forked the survey but cannot work before this weekend. Feel free to involve anyone else.
Amelia
@apottr
May 18 2017 15:55
@satenndrra there are also things like https://www.data.gov/ and https://data.world/ and http://archive.ics.uci.edu/ml/
Eric Leung
@erictleung
May 18 2017 18:11
A series of lectures, "The BD2K Guide to the Fundamentals of Data Science Series", given by the National Institutes of Health (NIH) in the US.
satenndrra
@satenndrra
May 18 2017 18:13
@evaristoc sure, count me in for the coder survey project. Need to work through some D3js, but I am ready to start whenever you folks want.
satenndrra
@satenndrra
May 18 2017 18:19
@apottr cool, thanks for sharing these resources!
CamperBot
@camperbot
May 18 2017 18:19
satenndrra sends brownie points to @apottr :sparkles: :thumbsup: :sparkles:
:star2: 2865 | @apottr |http://www.freecodecamp.com/apottr
evaristoc
@evaristoc
May 18 2017 19:38
Oh!! Great @PunkDado! Sorry...
evaristoc
@evaristoc
May 18 2017 19:52
@PunkDado I sent you a PM.
@satenndrra Excellent! We are trying to generate visualizations in d3.js for this project indeed. I will try to give you some opinions? Let me think? You can of course decide what kind of thing you want to do at the end. We can then here help you to polish it.
I will contact you soon.
jahala
@jahala
May 18 2017 20:39
Hi! Does this channel allow for some (very) general questions about data science / forecasting? Or is it strictly freecodecamp?
(perhaps someone could suggest another channel?)
evaristoc
@evaristoc
May 18 2017 20:53
@jahala it is mostly data-related topics, so any question regarding that that we could answer is always welcome!
Eric Leung
@erictleung
May 18 2017 21:03
@jahala feel free to ask general data science questions, we're a friendly bunch :smile:
jahala
@jahala
May 18 2017 21:04
Great, thanks! - I'm a total newbie in data science- so I just wanted to get some pointers before I dive in.
I have a dataset consisting of three columns - datetime, # of positions, position names. The data is a "snapshot" of every 30 minutes, for 24hours. I have about two years worth of data. You could say it is something like data about how many, and which cash-registers have been open in a store.
I came across Facebook labs "prophet" forecasting tool, which seemed "simple" enough to test for me:
https://github.com/facebookincubator/prophet
It doesn't support data that is as "fine grained" like mine straight out of the packet - but a user has added a pull request for "daily timeseries data" - which I've grabbed
So my question.. When dealing with data that is just integers - and which has a lower "bounds(?)" (never negative value) .. Is something like that forecasting tool the right one for the job?
Should I try a completely different direction?
jahala
@jahala
May 18 2017 21:13
What I want to try to get some insight into is seasonality and trying to forecast about a half year in advance
Also - If at some point it is combinable with number of visitors, try to estimate how many positions should be open (2-4 hours in advance) .. My data isn't really a store - we have some info about expected # of visitors.. I'm guessing this is more "ML" territory?
I was able to get some great results with just monthly or weekly numbers with "fbprophet", and some interesting findings about seasonality, but it doesn't really get where we need to go in regards to forecasting. And it very often forecasted a negative trend (to the point of going into negative digits) - even though not a single of our data values are less than 2
jahala
@jahala
May 18 2017 21:19
Anyhow, any kind of pointers or tips on things to read that may be related to this kind of data, and how to process it correctly for the kind of outcome I'd like is highly appreciated :)
Eric Leung
@erictleung
May 18 2017 22:01
@jahala great of you for sharing! Unfortunately, I don't deal with time series or forecast data so I can't help very much here. If you're going to use R, you can take a look at this CRAN Task View on time series analysis. There's a lot of tools there (and objectively overwhelming when you look at it), but it gives you an idea of what is possible.
jahala
@jahala
May 18 2017 22:03
Hi @erictleung - So far I've been using python. But I'm equally unfamiliar with both, so I'll have a look - thank you!
CamperBot
@camperbot
May 18 2017 22:03
jahala sends brownie points to @erictleung :sparkles: :thumbsup: :sparkles:
:cookie: 504 | @erictleung |http://www.freecodecamp.com/erictleung