In this project, we aim to wether a user will click on the ad or not. By applying data preprocessing techniques, conducting exploratory data analysis, and building classification models, we strive to create accurate predictions.
During the exploratory data analysis phase, we delved into the dataset through both univariate and bivariate analysis. We visualized distributions, relationships between variables, and identified patterns that could provide insights for modeling.
We implemented several classification models:
- Logistic Regression
- Support Vector Classifier
- KNN Classifier
- Decision Tree
- Voting Classifier
- Random Forest Classifier
- XGBoost Classifier
- AdaBoost Classifier
- Each model was trained and evaluated to determine its performance. The best estimator is Support Vector Classifier with f1 score of 0.967 on training set and 0.98 on test set.
- Daily internet usage and Daily time spent on site have very strong negative correlation with the target.
- Age and Area income have moderate positive correlation with the target.
- Gender has no effect on the target.
To view the notebook on kaggle please follow the link: https://www.kaggle.com/code/omarmostafataha/ad-click-prediction-98-f1-score