Citizens Bank Housing Price Predictor
by Bharat Gadiraju, Avery Peterson, Kenneth See, and Nakarit Suthapreda
For the 2019 Brown Datathon, we decided to take the Citizens Bank housing data from MA, PA, and RI over the last 13 years in order to train an artificial neural net that could predict a house's sale price within 10% of the actual sale price.
To do this, we used Python's Pandas library to clean and scrub the data into useful categories and quantities. Once this was done, we used the TensorFlow and ScikitLearn libraries in order to analyze the data on Jupyter Notebook and build the machine learning model.
1) Exploratory Data Analysis
- Used visualization and data manipulation to gage distribution of variables
2) Identify Relevant Data
- Determined which variables were relevant for our model
- Split into training and testing datasets
3) Clean and Scrub Data
- Transformed data into useful information
4) Binarize Categorical Data
- Transformed categorical data into a binary matrix for the model
5) Trained Neural Net
- Fit data into artificial neural network model