| Abstract [eng] |
The main goal of this work – to improve sales forecasting, applying business intelligence techniques. Improvement can be understood as facilitation of forecasting process, so to make it available to business people without deep mathematical analysis knowledge. The accuracy of forecasting results depends on completeness of analysis of current forecasting methods and their features while using them for sales forecasting. Research results are focused on data analysts, to help them make decisions about upcoming future and shift company‘s processes in correct way. Research helped to create data warehouse metamodel, which allows dynamic SQL queries generation, while using star data warehouse schema. Experiment results showed that time series models provided the most accurate results while forecasting pharmaceutical drug‘s sales, artificial neural network and linear regression models forecasted with larger errors and could not reach accuracy of time series models. |
