Using of Machine Learning for Predicting Advertising Campaigns on the Internet

The algorithm for efficiency of Internet promotion with applying of machine learning is suggested and tested in the paper. The object of the paper is the evaluating of Internet promotion efficiency, the subject is applying of data analysis tools in Internet analytics. The aim of the paper is to suggest and test the algorithm for prediction the efficiency of the Internet promotion. The paper is divided into 3 chapters, the first is about theoretical aspects of the machine learning (its history, notion, classification and algorithms) and the characterization of actual instruments for collecting, analysis and data processing. Milestones of machine learning development are pointed, advantages and drawbacks of using algorithms are compared. Likewise, the detailed description of modern web-analytics systems, their advantages and disadvantages and loading data through API is presented. The emphasis is on KNIME analytics platform that is used for the further analysis. The direct loading and processing of the data and prediction of Internet store orders is considered in the second chapter. The sequence of actions for loading data from Logs API of Metrika is described in detail. Based on the comparison of the algorithms’ efficiency optimal options for their usage in Internet marketing are suggested. In the last chapter options for practical used of suggested algorithm in management are proposed. In particular, possibilities of their interpretation, usage for prediction and creation of custom analytics systems for the particular business are considered.