Power transformer is one of the most important apparatus in the power system. Incipient faults in transformer can degrade both oil and cellulose insulation, leading to the formation of dissolved gases. Partial discharge (PD) is regarded as a cause as well as an indicator of the insulation degradation. This paper presents a study on the correlation between oil dissolved gas generation and cavity discharges by correlation-regression analysis. Phase resolved partial discharge pattern (PRPD) and principle component analysis (PCA) have been employed in the feature parameter extraction. Multiple regression equations are established, which could provide a reference for gas prediction by partial discharge characteristic parameters.