MODELING AND FORECASTING DEMAND FOR A DIGITAL PRODUCT

Different approaches to modeling and forecasting the demand for a digital product, namely paid activities of Facebook, are explored in article. The company is given reject its arrivals main form advertisements. available range data on company's allowed build forecast models based adaptive short-term methods, Brown method multiplicative Holt-Winters model taking into account quarterly seasonal factor. These have ability continuously take evolution dynamic characteristics studied processes, adapt these dynamics, giving weight high information value observations, if they close current time. were tested adequacy using number criteria, including RS-test, series criterion median sample, Student's t-test Darbin-Watson test. comparative analysis obtained results by choose that gives fairly accurate result. also showed there seasonality and, accordingly, significant decline at beginning year income growth recent quarters. 4th quarter next period built. compared with real prospects development products Ukraine determined, particular, spread use services many areas.