Users in web perform different actions. This can start from web page access to a transaction. Web service is a standardized way of communicating information between client and the server. Published services made available on web for the user to access the service. Interaction with web service can be considered at web service users or at web service state level.The type of transaction can be complete or incomplete due to normal actions or abnormal activities. The paper aims at discussing the website and web service user access patterns using Machine Learning considering only web service users. This will allow for further categorizing the users into specific levels. Learning of actions to be understood as anomaly can also be an outcome of the analysis. The dataset for website used for analysis is attributes used in KDDcup 2005 dataset and personalized web user dataset challenge. The dataset used for web service user access is based on the service assessment by several users on cloud. The paper aims at classifying the users based on query. The results are presented for analysis. The analysis can be used to understand behavior of users in organization and institution. The performance parameters analyzed are precision, recall, Accuracy, F1-score. © 2019, Institute of Advanced Scientific Research, Inc.. All rights reserved.