A Comparative Analysis of Sampling Techniques for Click-Through Rate Prediction in Native Advertising

Native advertising is a popular form of online advertisements that has similar styles and functions with the native content displayed on platforms, such as news, sports social websites. It can better capture users’ attention, they have gained increasing popularity in many platforms among advertisers. In advertising, Click Trough Rate (CTR) prediction essential but challenging due to data sparsity: non-clicks constitute most data, whereas clicks significantly smaller portion. The performance 19 class imbalance approaches compared this study use four traditional classifiers, determine effective methods for our ads dataset. used real traffic from Finland over course seven days provided by platform ReadPeak. resampling include undersampling techniques, oversampling hybrid sampling ensemble systems. findings demonstrate learning enhance model’s capacity classification much 20%. general, more stable comparatively. But, performed best Random Forest. Our also demonstrates ratio plays an important role model features importance.