Accurately forecasting future stock prices is an essential and important issue for an investor to make an expected profit, as well as decrease the investment risk. In the past, the topic that determines the factors crucial to the forecasting of future stock prices is rarely addressed. In this study, the support vector regression and TwoStep cluster analysis are integrated to propose an approach for tackling the stock price forecasting problems with a function of feature selection. The feasibility and effectiveness of the proposed method are validated through making a case study on three different levels’ indices that include the TAIEX, FTSE TWSE Taiwan 50 Index, and Taiwan 2303 Stock. The experimental results show that the feature selection mechanism can efficiently screen out the critical technical indicators for forecasting the future stock prices, as well as can remove the superfluous indicators that might interfere the forecasting abilities of other critical technical indicators. Next, the feature selection procedure can greatly reduce the total number of technical indicators to prevent the over-fitting of training the training data. The selected best simplified forecasting model cannot always provide better forecasting performance. However, the investors can only pay attention to less technical indicators, and can obtain a satisfactory forecasting result with a high accuracy. Hence, an investor can save time, cost, and effort while building a forecasting model, thus make more concentration on fewer financial analysis and trading strategies.

The link between the lean orientation and the agile model is only possible through the Supply Chain interface (SC) and the DC (Demand Chain), called "leagility." What does this concept cover? How is it achieved? The answer to this second question suggests a multidimensional repository based on the triptych: efficiency, efficacy, and responsiveness, the combination of which allows us to define not only the three determinants but also the related Key Performance Indicators (KPIs). It requires the use of instruments such as the Advanced Available To Promise Mode Batch (AATP MB / Time Real) to achieve a compromise between the objectives of the SC and those of the DC.