Click-Through Rate Estimation Using CHAID Classification Tree Model

Abstract

Click-Through Rate (CTR) is referred to as the number of clicks on a particular advertisement as compared to the number of impressions on it. It is an important measure to find the effectiveness of any online advertising campaign. The effectiveness of online advertisements through calculations of ROI can be done through the measurement of CTR. There are multiple ways of detecting CTR in past; however, this study focuses on machine learning based classification model. Important parameters are judged on the basis of user behavior toward online ads and CHAID tree model is used to classify the pattern for successful and unsuccessful clicks. The model is implemented using SPSS version 21.0. The dataset used for the testing has been taken from Kaggle website as the data is from anonymous company’s ad campaign given to Kaggle for research purpose. A total of 83.8% accuracy is reported for the classification model used. This implies that CHAID can be used for less critical problems where very high stakes are not involved. This study is useful for online marketers and analytics professionals for assessing the CHAID model’s performance in online advertising world.