Year of Graduation
Model of Dependence between Company's Profits and Advertisement Spend
The present work is devoted to the study of the influence of various factors on the costs of marketing on the Internet (on Google AdWords) for certain companies. The goal of the study is to simplify quarterly and annual planning in Google. Tasks that were performed to achieve this goal: Various sources were studied that contain information on the use of different methods in order to predict future values based on historical data, as well as on the basis of other parameters within the period; The methods which are necessary for the decision of the put purpose are defined and described; A set of data has been created from both internal and external sources; Identified and indexed data received in order to comply with Google's internal privacy standards; Determined the factors that affect the company's costs for Google Ad Network and their degree of influence; The values of the parameters that were revealed in the previous paragraph are estimated next year; The value of the target variable (costs) is estimated in the next year; Estimated distribution of advertising costs depending on the quarter (seasonal coefficients); The accuracy of the obtained model is estimated and compared with other ways of predicting costs (autoregression and time series). As an empirical basis for this work, various academic sources were used, devoted to the application of several methods of machine learning (linear regression, autoregression and time series) for solving various problems. The final implementation contains the use of each of the listed methods and a comparison between them. The result of this study is three models built, one of which is currently used to predict the costs of customers in Google and shows a fairly high accuracy.