Sunday, 9 November 2014

CTR Prediction in Machine Learning

1. Concepts:Click: the number of times, among the above impressions, the user (UserID) clicked the ad (AdID).Impression: Number of times the Ad was displayed in the session.click-through rate(CTR) = #clicks / #impressions
2. Statistics:
ROC/AUC, MSE(mean squared error), RMSE(root mean squared error) Weighted mean,
Logistic regression, Normal/Gauss distribution, T/Z-test and Confidence interval, K-fold
cross validation, Factor Analysis, (Stochastic) Gradient Decent.

3. Data Mining:
Linear  classifiers, Supervised Classification, Logistic Regression, Naive Bayes, SVM, Perceptron, PA(Passive Aggressive),EM algorithm, Ensemble Learning, AdaBoost.

Reference:
http://mlwave.com/predicting-click-through-rates-with-online-machine-learning/
https://docs.google.com/document/d/1oheEV8w0JphWooAtfcgQNrzFz7y68XFrfKhr1E14kwo/edit
http://sjsubigdata.wordpress.com/2014/02/09/predict-the-click-through-rate-of-ads-given-the-query-and-user-information/#_Toc387689969

No comments:

Post a Comment