Predicting Ad Click-Through Rate
Optimizing digital ad performance with machine learning in R.
Result: A Bayesian-tuned XGBoost model had the lowest RMSE of all models tested. It also identified the ad factors that drive engagement.
Goal
Predict click-through rate (CTR) and identify the ad factors that drive engagement, such as quality, relevance and audience targeting. The findings help businesses allocate ad budgets more effectively.
Approach
- Preprocessing: Cleaned the data, handled missing values and encoded categorical variables.
- Feature engineering: Identified the key predictors of CTR.
- Modeling: Compared Logistic Regression, Decision Tree, Random Forest and XGBoost.
- Tuning: Used Bayesian Optimization for hyperparameters and compared models on RMSE.
Takeaways
XGBoost captured complex, non-linear ad-performance patterns better than the other models. The final model gives marketers a clear basis for refining targeting and getting more return from their ad spend.