This repository contains two neural network implementations that I wrote for an undergraduate course on neural nets. One is a multilayer perceptron (MLP) and the other is a radial basis function network (RBFN). The aim was to use data on houses in Boston, USA, to predict their sale price. There we 13 fields of training data as well the house price to test against.
The RBFN worked better, with an average error of about $2500, which is about a 5% error (the average house price in the data set was around $50k).
This project helped me improve my Python skills. It was a difficult project given the time we had to complete it. It taught me about working hard to meet a deadline, and about how to manage time when you have to wait for hours for experiments to run.