: Nowadays, the volume of data rises with a considerable rate. Thus, there is an indelible demand in techniques able to analysis of the mentioned data and intelligently discover the hided knowledge in them. Data mining provides some tools for analyzing mass data bases, discovering the processes, patterns and knowledge, using this data and information. The Sepahan Oil Company with the production capacity of more than 450000 Ton of various kinds of oil, is one of key industries in Iran. Considering its rich data base, Sepahan Oil Company is a perfect target to applicate the data mining techniques and consequent discovering of knowledge embedded in such data base. The efficiency and qualitative parameters of base oil as the main production of this company are considered as two of the most important factors affecting the final quality of products. Incompatibility of the of these parameters with the desired parameters represented by product specialists is one of the basic challenges in this production line. Current thesis, accordingly, is pioneer at manipulating data mining techniques to process the data bases available in Sepahan Oil Company and consequent prediction of output parameters in base oil production line. Coming across with multiplicity of variable items during data analysis and encountering dependence of this items on each other which make some difficulties at analyzing them, this investigation is intended to lower such problems by declining the variable size and eliminating their dependence using principal component analysis. In order to meet this target, intelligent techniques of knowledge discovery such as neural networks, regression trees, support vector machines and conventional regression methods have primarily been described and used to predict the output parameters of production of base oil after the data selection and data preparation. Subsequently, Principal factors resulted from principal component analysis method have been assumed as the input of these methods to evaluate the impact of such factors on data mining techniques as well as comparing its outcomes with implementation results of data mining techniques on basic data. Based on received evidences, it is fair to say that principal component analysis method represent a better function when accompanied with data mining techniques and lead to improve them. Drawing a comparison between distinguished methods, neural network method based on principal factors has been considered as the best method at prediction of output process parameters. Finally, affecting method of input factors on target parameters has been determined using the regression trees.