Recommendation systems (RS) are used by many businesses to identify product recommendations made by consumers that interact with e-commerce websites. Massive growth in both commodities and consumers has recently encountered significant difficulties. Many websites overwhelm the user with alternatives at once, which causes a lot of confusion. Additionally, a crucial component of RS is locating the appropriate product or active user. Based on consumer preferences and sociodemographic trends, products are already recommended. In order to enhance the user behaviour matrix, a hybrid-actionrelated recommendation based on K-Nearest Neighbor Similarity (HAR-KNN) combines the simplicity of hybrid filtering with the creation of feature vectors. It uses both quality and quantity classifiers to categorise properties. The suggested methodology also addresses weaknesses in past methods of feature analysis and user preference evaluation. The SOM AND KNN classification technique has been authorised for the purpose of locating data about user behaviour online and in real time for a particular user group that contains a huge amount of data in connection to the commonalities among many users and target users. Highly predictive metrics such as Precision (P), Recall (R), and F, as well as Mean Absolute Error (MAE), Mean Square Error (MSE), and Root Mean Square Error, are used to assess a test result (RMSE).