Skip to content

goldentom42/kaggle_porto_seguro_2017

Repository files navigation

Porto Seguro's Safe Driver Challenge hosted by Kaggle

Introduction

In late 2017, Kaggle and Porto Seguro, one of Brazil’s largest auto and homeowner insurance companies, organized a competition where kagglers were challenged to build a model that predicts the probability that a driver will initiate an auto insurance claim in the next year. While Porto Seguro had used machine learning for the past 20 years, they were looking to Kaggle’s machine learning community to explore new, more powerful methods. A more accurate prediction will allow them to further tailor their prices, and hopefully make auto insurance coverage more accessible to more drivers.

Models

In this competition I managed to finish at 33rd rank over 5000+ teams. My solution was based on the following set of models:

Most of these models used a subset of the dataset features whose selection was performed using my py_ml_utils/feature_selector package

Stacking was done using the linear stacker available here

How to build the solution

Simply clone the repository and run ./build_solution.sh

Dependencies

About

No description, website, or topics provided.

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published