Internet plays an important role in our day-to-day lives, and we are using many e-commerce websites. Cyber-criminals are using phishing attack technique to steal user's sensitive information like personal details, and bank details. It may lead to loss of both client and original seller. Our work involves the decision tree and best performing classifier to identify the significant dimension. The entropy features include SFH, pop up window, and URL of anchor. Through which, RF with 5 features at each split achieves 84.8% accuracy. In our work, RF with 6 features achieves 96.30% accuracy. However, the RF with all features performs only 93.20% accuracy. © 2019 IEEE.