Bluebook for Bulldozers

A heavy machinery price predictor

Overview

Bluebook for Bulldozers was a kaggle competition in 2013. I used the fastai library to analyse the sales data for heavy machinery. I used various methodologies to clean the data while preprocessing, used statistical tools to extract information from the given and then extrapolated dataset, used feature importance, correlations and hyperparameter tuning and finally the random forest regressor to generate a model that could accurately predict the price of the equipment. To generate the model, three datasets i.e., training, validation and test set were created out of the initial raw dataset.The model error score was in-line with top 20 of the kaggle leaderboard.

Technology used

Python, Pandas, Fastai, Numpy