Abstract—Improvements in technology have led to an increasing number of internet users which support cyber-attack growth. Phishing attacks have multiplied rapidly since 2021 according to the Anti-Phishing Working Group (APWG). The time to identify legit and phishing sites quite take a long time in this fast-paced era. The difficulties in identifying phishing sites visually without a deep check can risk users and organizations for their security. To address this issue, automation using machine learning to speed up phishing site classifications is proposed. This study focused on developing phishing site detection using a machine learning approach. Specifically on Random Forest (RF), Extreme Gradient Boosting (XGB), Light Gradient Boosting Machine (LGBM), and Multi-Layer Perceptron (MLP) neural networks. Those algorithms are tested on three different datasets to evaluate the impact of feature quantity and dataset size towards model performance. Experimental results show the highest accuracy of 98.77% using XGB in the first dataset. Meanwhile, RF can achieve accuracy values of 98.2%.