Ricoh has developed a new technique for forecasting volumes of used products to be collected from the market. The extraction from Ricoh customer databases of factors useful in forecasting, such as numbers of employees and numbers of copies used, and the analysis and accumulation of collection volume trends for each product has made possible collection forecasts that are very similar to the actual results. Furthermore, reflecting actual collection results in successive forecasts and correcting errors, we are able to increase the accuracy of forecast values over time. The application of this used product collection forecasting technique enables detailed collection forecasting by area, by period (month, half-year, year), and by number of copies used for each model and makes it possible to formulate appropriate reconditioned machine production and sales plans based on forecast values.
Figure 1: Example of Collection Volume Forecast
This system makes it possible to forecast with extremely high accuracy even collection volumes several years into the future by discovering mathematical rules concerning product collection and autonomously correcting those rules on the basis of newly accumulated data. In volatile market conditions, discovering statistically significant rule comes with difficulty. In the difficult situation, this system can forecast stably with high accuracy by using time-series analysis and statistical modle. Therefore, this system can forecast accurately both next month and years after.