In smart homes, different sensors and devices are to interconnect and communicate over the Internet. The information from various sensors in different locations has been gathered, to identify and understand the daily activities of residents in a smart home. Any irregularities can compromise the overall security of the home. For this reason, we explore methods to detect activities for secure monitoring in smart homes, using various machine learning (ML) techniques to classify activity data from users and devices, which will help improve home security. We used techniques like the Random Forest, support vector machine, K-nearest neighbor, logistic regression, and Naive Bayes theorem to build a baseline model based on the user’s and device’s normal behavior routine sequence and identify the activities. Experimental analysis using a real dataset demonstrates that the k-nearest neighbor algorithm achieved a high quality in classifying activities, outperforming other algorithms in comparison.