These are chat archives for FreeCodeCamp/DataScience

1st
Aug 2017
Timothy Javins
@timjavins
Aug 01 2017 00:38
Today in Forbes Magazine/website:
Eric Leung
@erictleung
Aug 01 2017 02:02
@timjavins thanks for sharing! Pretty much all the careers that were doing "data science" before it was cool :laughing:
CamperBot
@camperbot
Aug 01 2017 02:02
erictleung sends brownie points to @timjavins :sparkles: :thumbsup: :sparkles:
:cookie: 130 | @timjavins |http://www.freecodecamp.com/timjavins
Harshit rathod
@harshitrathod
Aug 01 2017 09:16
Hello does anyone setup ML workspace on google cloud?
Harshit rathod
@harshitrathod
Aug 01 2017 17:52
@all I have one doubt about missing value. suppose in my model I have variable with missing value. to resolve this issue I have filled data with mean values. Now this model is predicting live data and suppose input has missing value, should I need to pre-process this like I did with training set? and what will happen/impact if I do not do this?
ErMochi
@ErMochi
Aug 01 2017 17:55
Hi all! I'm new here but in my opinion, yes, you should do the same...
Another question, maybe I would take the median instead the mean, is more stable when you have enough data
Eric Leung
@erictleung
Aug 01 2017 20:52
@harshitrathod sorry, I've never setup a workspace Google Cloud. And median sounds better as @ErMochi has mentioned because it is more stable, especially when you're getting more data. I would caution that your predictions are only as good as your data. And if you're essentially imputing missing data, I'd be mindful of how you interpret your results.