By using smartphone sensors human activity recognition is an active field of research. A fast-growing subject of machine learning in the artificial intelligence field is Human Activity Recognition (HAR) because it makes life easier and more comfortable. Extensive research has been conducted on simple human activity recognition but very small research has been conducted on complex human activity recognition. Humans carry out a variety of activities during the day, the majority of which are complex. For the recognition of complex human activities different sensors are used to achieve high accuracy. This study assesses six classification algorithms in two scenarios: one without accelerometer data and one with it. Naïve Bayes, BayesNet, Logistic Regression, Decision Table, Random Forest, and Lazy IBK were evaluated for accuracy, precision, and recall. In the absence of accelerometer data, all algorithms showed competitive accuracy. Lazy IBK stood out with high precision and recall. With accelerometer data, all algorithms improved accuracy, with Lazy IBK reaching 100%. In the future, the study plans to develop an Android app for data collection and recognize complex human activities using accelerometer data. This research highlights the role of data sources in algorithm performance and Lazy IBK’s suitability for activity classification.