Prediction of the Target User Segment for Paid Advertisement

The primary objective of this work is to develop a methodology for identification of a segment of customers to whom advertisement(s) of a particular product can be shown. One of the important aspects of this work is data. To create a database of the customer, datasets from different sources/websites are collected, cleaned and preprocessed to prepare the customer database. DBSCAN (Density-based spatial clustering of applications with noise) clustering algorithm is then used to form customer segments. KNN (k-Nearest Neighbour) algorithm and performance metrics like Silhouette Coefficient are used for the smooth conduct and working of the clustering algorithm. The customer database and the labels generated after clustering are used to build and train a Logistic Regression classification Model. For identifying the customer segment for a particular product, the characteristics of a product are converted into a database by doing a Cartesian product. The classification model is then used to predict the classes/labels for the product. The labels predicted by the Classification Model are used to extract a target customer segment(s).