Omorhefere Imoloame Undergraduate Dissertation 2016/17

Predict Someone's Income

Supervised by E.Vasilaki

Abstract

The ability to predict someone’s income is a valuable asset

This piece of work aims at building parametric and non-parametric classifiers on the Adult dataset that predict someone’s income to a high accuracy. This project also explores several strategies to improve classifier accuracy such as ensembles, PCA and feature engineering. A voting classifier has been able to attain an accuracy of 84.57%. Finally, further work to be done to improve accuracy is discussed