| Theme | Industry 4.0 |
|---|---|
| Project title | Forecasting model for the prognosis of short and medium-term sales by search engine data (ProSuma) |
| Project duration | 01.06.2016 – 31.05.2018 |
| Download | |
| Press release |
The planning of delivery dates and sales volumes presents a challenge in the enterprise logistics planning. This planning is complicated by hardly forecastable sales fluctuations for example caused by promotion. Deficiently forecasted sales lead to supply shortages and with it to poor logistics performance, whereas overestimated sales cause increased logistics costs due to inventory and capital commitment. Conventional forecasting methods show deficits in countering this uncertainty because of low data timeliness, low level of detail and extensive effort.
The aim of the research project is to develop a forecast model for sales volumes on specified product level based on search engine data. It is expected that, in comparison to conventional forecast models, the forecast error can be reduced as well as the forecast horizon can be increased. Overall, the central issue is to ascertain whether and to what extent the logistics performance can be improved by search engine data-based forecast.
Publications about the project
In the production of stock goods, manufacturing companies are faced with uncertain customer demand. In order to counter uncertainties, an increased inventory is necessary in order to be able to meet customer demand. The costs incurred are influenced by the ordering behaviour given the forecast uncertainty. Ordering behaviour is largely determined by the ordering policy. Therefore, the influence of forecast uncertainty and ordering policy on the resulting storage costs was investigated by means of sensitivity analyses. Accordingly, forecast uncertainties require larger inventories under the (t, S) policy than under the (s, q) policy.
stock planning, ordering policy, forecasting
Combining standard time series models with search query data can be helpful in predicting sales. We include the search volume of company as well as product-related keywords provided by Google Trends as new predictors in models to forecast sales on a product level. Using weekly data from January 2015 to December 2016 of two products of the audio company Sennheiser we find evidence that using Google Trends data can enhance the prediction performance of conventional models.
Google econometrics, forecasting, search query data
The forecast of sales volumes represents a challenge for the production planning. Above all, sales forecasts that are difficult to predict, such as those caused by promotions, are obstructive. Often, additional information from macroeconomic indexes is not topical, the level of detail of products to be forecast too low and the forecast expenditure too high. Aim of a research project therefore is to develop a model based on search engine data to forecast sales volumes at product level. By the use of complementary application of search engine data to the sales forecast is expected that the forecast mistake can be reduced compared with conventional forecast models upon product level. In general it should be clarified whether and in which extend the logistical efficiency of an enterprise can be improved by search engine data based forecast of sales volumes in the production planning.
production planning, sales forecast, search engine data, forecast model
