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
@chessybo I think your analysis is clear and shows in different graphs there is a difference in attendance according to demographics.
@arunkumar413 A first usual step is segmentation using descriptive statistics and possibly unsupervised learning techniques. Although not inherently predictive, it could become predictive if the description holds for every case.
If you want to apply supervised ML for that, a simple recommended approach that I saw being applied in a now closed kaggle competition was a regularized logistic regression with hash encoding: https://www.kaggle.com/c/avazu-ctr-prediction/discussion/10927#58054. There were variations and improvements of this same method by several participants. You can go through the whole discussion of the aforementioned competition to see if you find something that meet your expectations.
In a different order of ideas, I have been contacted by an NGO here in Amsterdam to help them with their digitization policy as a volunteer.
The NGO is veeery behind, basically run by either old people or with poorly IT-literacy, so there could be a lot of work to do. However there are a few young members looking for modernising the organisation. For what I have seen, their IT needs are plenty and it could happen that they might need some new applications, possibly a lot of of hard, extensive work.
I have thought at fCC as possible provider. Question to you: Is fCC moving from attending NGOs to rather developing modules of more general use and impact, or is still fCC working closely with NGOs in partially/fully customised solutions?
Hope to hear from you.