Year of Graduation
A Model of Classification and Forecasting of Companies' Expenditures on Contextual Advertising
This paper studies the main strategic patterns in contextual advertisers’ behaviour. Companies’ behaviour is seen in terms of the dynamics of expenditures on the specified type of internet marketing. The main hypothesis of this study is the ability to explain the level of advertisers’ expenses on advertising through a predetermined pattern. The main purpose of research is to solve the problem of customers prioritizing through their classification. There is a significant difference in behaviour of those who are interested in the long-term execution of contextual advertising and those who use it as a temporary sales promotion or branding tool. Machine learning and neural classification network technologies are used to carry out clustering and classification of clients of one of the largest contextual advertising placement services. The main expected result is a model that interprets input signals to understand whether the advertiser is interested in long-term collaboration. This paper is divided into three main sections. The first section gives an overview of the background, problem statement and analyses the preliminary works. Analysis methodology is described in the second section. Some conclusions of the research are drawn in the final section. The results of clustering and the development of the predictive model are described in the final part of the paper.