Click-through-rate is always a key metric for companies to evaluate the effectiveness of advertising. This project uses Logistic Regression, XGboost, and LightGBM to predict whether the user with particular sets of features will click the advertisement or not.
This projects aims for practicing using two newly learned ML methods: XGboost and LightGBM. In the code, I show why these two methods has a higher performance score than traditional Logistic Regression, and what is the advantages and disadvantages of XGboost and LightGBM
Data is download from Kaggle CTR competition. Since this project meaning focus on practicing new ML methods, I only randomly select 1% of the original data (still around 400,000 data). This 1% data can be accessed from here.