Transformer failure could occur in terms of tripping that results in an unplanned or unseen outage. A good maintenance strategy is therefore an essential component in a power system to prevent unexpected failures. In this paper, the causes of transformer failure within the power transformer systems have been reviewed. Data is obtained from the transmission substation assets from the whole of Peninsular Malaysia for the past 5 years. However, the challenge is that the problem descriptions of the datasets are all in text formats. Thus, text mining approach is chosen for the data analysis using R. This paper covers the most common steps in R, from data preparation to analysis, and visualization through wordcloud generation. This study mainly focuses on bag-of-word text analysis approaches, which means that only word frequencies per text are used and word positions are ignored. Although this simplifies text content dramatically, research and many applications in the real world show that word frequencies alone contain adequate information for many types of analysis. As a result of analysis, keywords like "leak", "lightning", "animal", "cable" and "temperature" are identified as the main causes of transformer failures based on the number of word frequency in the tripping dataset. Further enhancement could be made in the future to predict the failure beforehand using predictive analytics approaches.