These are chat archives for FreeCodeCamp/DataScience
discussion on how we can use statistical methods to measure and improve the efficacy of http://freeCodeCamp.com
This is a VERY interesting conversation I brought here a couple of times but never discussed properly: The possible end of the Data Science Bubble and the Auto-ML systems.
Not that I think we are close to those days but it is also naïve to think that the bubble will last forever. And yes: there are now many solutions put forward by different online providers that will make the whole calculation for you.
This is unfortunately a natural steps in all industry that become mature and more for those industries based on computers: to centralize (eg. moving all ML activities to India) and to automate.
Additionally, thanks to things like Kaggle (and freeCodeCamp...) there is an increased supply of labour force and algorithms that will sooner or later affect the job availability and the wages.
Just to complement that: in many meetups (Nederlands) I noticed how many small companies that were servicing the data analysis market were moving to:
Not enough work, they argued. Much of the standard analytical services are now offered by the big players.
Hurry up to get there. And even if you get there, be prepared to last in that job for few years. Specially if what you are looking for is a "secure job".
The opportunities are actually in the specializations and the added value, not in the things that are becoming commodity.
@Mps24-7uk I haven't tried any pickling from Heroku but it looks like either a version or the
cPickle package are not installed there? I think you should be able to see what it is available and if you can install a package. I would suggest to also check the Python version.
Probably the best way to deal with this is using containers or any other tool that will port your settings into Heroku. Not my area of expertise though.