Predict sales prices and practice feature engineering.
This is from a Kaggle competition.
https://www.kaggle.com/c/house-prices-advanced-regression-techniques/
Unzip the .zip file. The only files you need are train.csv and data_description.txt. Do not use the test.csv as it is only useful for the Kaggle competition.
You will not be submitting results to Kaggle. This is only for in-class work.
Fork and Clone this repository.
You are trying to predict the Sale Price of a house based on various predictors.
This is a regression, so after preparing the the data (eda, splits, feature engineering, etc.), do a linear regression first using the sklearn implementation. What is the MSE on the test set, not on training.
Next, build a neural network. You are free to try as many architectures as you'd like, also try various optimizers, and other miscellaneous hyperparameters.
Plot the MSE for both training and validation as a function of epochs.
Once again, what is the MSE of this model on the test set.
Which model performed the best?