final team project 2

train

test

Weight, payer code, max glucose, a1c res and medical specialty contain too many missing values. Diag_1, Diag_2 and Diag_3 have too many different values which could lead to hundred of small dummy variables which would eventually lead to overfitting.

Changes ? values in race to Unknown to make results more readable.

Changes ? values in race to Unknown to make results more readable.

test predictor

Test predictor.

Changes target to nominal variable to allow for logistic regression training. Also changes admission source and type id to a nominal variable as order should not matter.

Normalizes large columns like num lab procedures and num medications to prevent bias in logistic regression as large values can lead to bias.

Ignored readmitted and encounter id columns when training to prevent bias

Training predictor

training predictor

Changes admission source and type to nominal variable