Final project for CS254 Machine Learning by Gabriela Salazar Lopez, Chongqing Gao, Lichao Yang, and Isabelle Francke

Summary: Code for creating and evaluating a machine learning model that predicts the severity of depression based on patient data. We used the 2019 NHIS adult interview dataset to train our model.

InitialDataExploration.pynb: exploring dataset, looking at size, missing data, datatypes, etc

inital_model_data_preprocessing.pynb: preprocessing/cleaning of data for training initial model

Data_Cleaning_Final.pynb: final data cleaning, removes irrelevant columns, removes missing data codes for numerical and categorical ordinal data without N/A's, replaces appropriate NaNs with an N/A code for categorical data, splits into training and testing data, normalizes numerical data, one-hot encodes nomical categorical data

DataImputation.ipynb: removes data randomly and imputes it using a simple imputation strategy (univariate imputation: mode for categorical data, median for numerical data)

FAMD.pynb: performs dimensionality reduction on features using FAMD

RandomForest.pynb: training for initial random forest model

RandomForestFinalModelParameterOptimization.ipynb: optimization and training for final random forest models including parameter search with cross-validation

RandomForestFeatureSelection.pynb: analyzes feature importance using permutation and iteratively selects features using cross validation. trains models with reduced feature datasets.

RFC_ImputedDataTest.pynb: Testing on all random forest models using imputed datasets.

Logistic_Regression.ipynb: training for initial logistic regression

SVM.ipynb: training for initial SVM model with binary encoding