Precise and Accurate Job Cycle Time Forecasting in a Wafer Fabrication Factory with a Fuzzy Data Mining Approach

Table 5

The proposed methodology compared to data mining.

Data mining steps	The proposed methodology
Application domain identification	Job cycle time forecasting for a wafer fabrication factory
Target dataset selection	The most recent PROMIS data of a major product type with normal priority
Data preprocessing	PCA, partial normalization
Data mining	(i) FCM for job classification; (ii) FBPN to model the relationship between the job cycle time and six job parameters
Knowledge extraction	Training data (3/4 of the collected data)
Knowledge application	Testing data (the remaining data)
Knowledge evaluation	Forecasting precision (the average range) and accuracy (MAE, MAPE, RMSE)

*PROMIS: production management information system; MAE: mean absolute error; MAPE: mean absolute percentage error.