<p>Dissolved gas analysis (DGA) is popular among power utilities for the detection of incipient faults in transformers. Various techniques are available that identifies the fault type using gas concentrations. However, only a handful of them consider factors like oil-filtration, which affect the concentration of these gasses and hence, the accuracy of detecting the incipient faults. The present work proposes a Principal Component Analysis based approach to analyse DGA data to track the temporal development of the fault point while taking into consideration the oil-filtration history. This will provide a better insight of the gasses being generated inside the transformer. A methodology has also been proposed later in the paper that can predict the type of fault accurately by mitigating the effect of oil-filtration on the concentration of the dissolved gasses. The proposed methodology can be readily applied to analyse DGA data and provides accuracy that is comparable to available Machine Learning based techniques. The fault identification accuracy of the proposed method was tested using DGA data obtained from IEEE Data port, local utility and IEC TC-10 data set. An accuracy of more than 95% is obtained from the proposed methodology corresponding to the fault types.</p>