Abstract:
Growth of the list of names of sold production has forced many trading organizations to search for new approaches in forecasting sales. It was possible to predict sales of tens kinds of the goods by the detailed analysis, but to make it with hundreds and furthermore in thousand numbers it is impossible. Thus for any company which are applying for the post in the market it is necessary to build forecasts of sales for everything, or even most widespread kinds of the goods, otherwise it would test significant financial losses. The basic characteristic feature of a similar sort of problems is presence of a database about sales of the big assortment of the goods (from hundreds units up to tens thousand). It means, that there is a plenty of time numbers of the sales demanding periodic lengthening for planning of the budget and acceptance of administrative decisions. Obviously, mistakes in forecasting demand conduct or to excessive filling a warehouse that raises costs on storage of production, or to impossibility to sell this or that goods to the buyer because of its absence in a warehouse that is negatively reflected in loyalty of the client to the company and reduces profit.
Clearly, that the company always starts with economic efficiency business-processes. Therefore till some time not quite effective forecasting of sales and planning of warehouse stocks can suit a management. However with toughening a competition between the companies, and also with expansion of assortment of production, qualitative and operative forecasting of demand becomes more and more demanded. Assortment of company Valmi Automotive - more than 100 000 positions of the goods, and branches of the company are practically in all regions of the country, therefore effective distribution of the goods on warehouses is an obvious problem.
Purposes are:
- reception of representations about a modern condition and tendencies of development of scientific and practical development in sphere of forecasting of the sales connected with growing assortment of the goods in points of wholesale trade;
- revealing of lacks and unresolved questions in available development, a choice of a direction of own researches;
- a writing of the software product realizing as much as possible correct decision of the put problem.
For achievement of objects in view is necessary to solve a number of following problems:
1. To lead the analysis and to choose factors, influencing on sales volumes of auto spare parts in the company.
2. To lead the statistical analysis of time numbers of sales of auto spare parts and to define their character (seasonal prevalence, autocorrelation, etc. parameters).
3. To lead selection of the most significant factors influencing sales volumes of auto spare parts by their statistical estimations.
4. By means of analytical packages to lead experiments on forecasting sales volumes with use of classical methods and neural networks.
5. To prove a choice of a method and to realize a program complex.
Scientific novelty and practical value:
Scientific novelty consists in application of the newest development for the decision of a problem of forecasting of sales, and also application of methods of forecasting for the decision of a problem of optimization of warehouse stocks. Novelty aspects contain following results of research:
- application of scientific approaches to the analysis that is new to the Ukrainian car business;
- automation of process of decision-making;
- application of neural networks for the decision of the task in view, to in exchange standard mathematical methods, not giving due result;
- applications of a problem of forecasting of sales with the purpose of filling of the warehouse, to in exchange standard methods of optimization of warehouse stocks.
The review of existing researches and development
Methods of forecasting of sales share on statistical and expert, based, accordingly, on processing of quantitative data and estimations of experts (figure 2.1). The majority of the companies when there is such opportunity, prefers to use the first type of forecasts, as like (in the certain conditions) it is much cheaper and easier than accuracy, than attraction of experts. Most simple and widespread from statistical methods - extrapolation. Its essence consists in use of the historical given previous periods (on sales volumes, warehouse stocks, quantity of buyers, etc.) For definition of the general tendency and it " prolongations in the future ". Thus all is meant, that the major factors influencing the investigated parameter, will continue the action in the foreseeable future and the certain tendency remains in force for the nearest period.
This method well approaches for forecasting such parameters as volume of the market, size of warehouse stocks, a level of development of manufacture, however concerning sales its application is limited mainly to the mature markets with the developed balance of forces and low probability of qualitative changes which could affect on situation. Any unexpectedness breaking a former course of things, sharply lowers accuracy of the forecasts based on extrapolation. As is known, always it is possible to tell with 75 percent probability, that tomorrow there will be a same weather as today, however at us behind a window rains are continually replaced by clear weather, and the winter is followed with spring. One more restriction imposed on application of extrapolation, consists in necessity of reception full and a trustworthy information about the last events.
For example, the retail trading companies are used for forecasting by the data received from scanners a stroke-codes on cash departments who with exhaustive accuracy inform, what goods have been sold and in what quantity.
There are the receptions raising accuracy of extrapolation at insufficiently stable conditions. In particular, one of them consists in use of the weighed parameters when data for last periods have greater value, than old. At presence of the big file of the information smoothing is applied to reduce influence of random factors and seasonal fluctuations. At all lacks extrapolated methods are easily automated and consequently are often applied when it is required to make quickly the big number of forecasts (for example, at a prediction of a sales volume in the large company selling production of hundreds and thousand of names). Besides extrapolation always can be used in parallel with other methods of forecasting, applying numerical numbers as the first approach. In particular, extrapolation can be combined with expert estimations. If in the market there are any serious changes (acceptance of amendments to the legislation, sharp reduction of price by the competitor, occurrence of qualitatively new goods in the same group, etc.), The companies address to experts that those have estimated influence of similar changes on market tendencies. As experts managers on sales of the firm which correct statistical forecasts in view of the knowledge of the market often act.
|
To the begining
|
