Year of Graduation
Modeling the demand function for "Becker" Retail Chain
School of Applied Mathematics and Information Science
One of the most important conditions for a shop to achieve the maximum profit is the equality of supply and demand for goods. Many companies engage in the collection of data on sale; however, they do not continue to process data. This paper presents an analysis of time series obtained from the sample data of a real retail chain company "Becker ". Data described the review received from the sale of eight thousand products during the year. To further explore all these goods it has been allocated to ten groups. During the analysis, basically, non-parametric methods were used. First of all, time series are smoothed by simple moving average method. Using the Brodsky-Darhovsky algorithm, sample is tested for the presence of change-points and depending on the results areas with different tendencies are allocated. For each interval, using the least squares method, one of the seven basic functions, which best describes the changes in data, is modeled. The criterion of choosing of the optimal function is the highest coefficient of determination. In addition, for each selected function the average error of approximation is calculated, and also for each selected interval weekly seasonality is calculated. ARIMA model, which best describes the analyzed time series, is created by the Box-Jenkins method. Then we make the forecast for a certain period of time. In the result of the analysis we suggested the model, which determines the demand function for goods depending on the time period of the year.