ml_stem_analysis

Create a public repo called ml_stem_analysis on github and share with us.

Data: https://data.world/education/2010-federal-stem-inventory/file/2010%20Federal%20STEM%20Education%20Inventory%20Data%20Set.xls

Stage 1:

1. Calculate % growth of funding between year 2008 & 2009.
2. If funding is positive, tag it as 1, if funding is negative tag it as 0. This is the target variable.

Stage 2:

1. Create graphs of univariate distribution of all non funding variables and share on a jupyter notebook. Just FYI - Funding FY2008, FY2009, FY2010 are the "funding variables"
2. Calculate mutual_info_score of target variable created in stage 1 & ALL non funding variables and share on a jupyter notebook.

Stage 3:

1. Divide data into train & test samples. (70-30 split)
2. Select features & build xgboost model. You will be judged on roc_auc_score on test sample.
3. Write testcases on all user defined functions using pytest framework. This is one of the most important steps of this interview.