Click-through rate prediction using decision tree

Abstract

Advertising click-through prediction is a cardinal machine learning problem in online advertising. Click-through prediction anticipates the prospect of an ad to be viewed by the user. It necessitates binary classification to stratify the ad into either categorical or numerical. The forecast is constructed on ensuing attributes: publisher, user and ad content statistics. The machine learning algorithms when furnished by pertinent data sets about users, advertisers and display platforms can be employed to foretell the tendency of a visitant on an online platform, to click on the ad presented [1]. The quantification of such proficiency is the click-through rate (CTR), which is the ratio of clicks on a specific ad to its total number of views [2]. A higher CTR indicates the pronounced proclivity of an ad and its triumph in the advertising industry. The prediction is rooted to the elementary questionnaire like, how does the mélange of ages influence a client’s choice towards a certain ad? What is the impact of people’s financial status and net worth in regulating the selections of ads? How does the plethora of people’s métiers and limited time factors govern the prediction? and so others. Decision tree classifier is the foundation of our paper, a tree-based algorithm which is a highly espoused algorithm in machine learning [3]. A recursive algorithm, which splits the nodes persistently into subsets by hinging on the antecedent observations. Decision tree classifier is a sequential diagram portraying all the expedient decisions and analogous outcomes. The implementation, coding and development of the algorithm are carried out in Python. The results witness improvement in the prediction by using decision tree algorithm.