ALLSTATE CLAIMS SEVERITY Overview: Allstate, the largest casualty insurer in America strives to improve its claim service and published a dataset which will help in building models that can predict claims severity. This case study aims to develop automated models for predicting the cost (severity) of a particular insurance claim. Dataset given to us contains data of accidents that happened to households (each row of the data corresponds to a particular household) with target variable as cost of the claim, which is numeric in metric. As data given to us is regarding information of individuals, Allstate did a good job in highly anonymizing the data (changed feature names) and this aspect of this dataset makes it really hard to understand features. Dataset contains claim records of Allstate clients. Entire dataset is highly anonymized.
Dataset: This data set and problem overview is hosted by Allstate on Kaggle. Here is the link to download the data. https://www.kaggle.com/c/allstate-claims-severity/overview/description