Automated residence depends on precise activity identification to offer personalized services, improve safety, and improve user experience. This research article compares the performance of CatBoost, a potent machine learning algorithm, in enhancing activity detection. To conduct the assessment, we construct an elaborate dataset using the OpenSHS for a particular user over a span of a time. The dataset captures a wide array of activities performed within an Automated residence environment. Using this dataset, we train and evaluate the CatBoost modelalongside other advanced algorithms. We analyse variousfactors such as accuracy, resilience, and computational efficiency to gauge CatBoost performance. The outcomes demonstrate that CatBoost outperforms other algorithms inprecisely recognizing activities in the Automatedresidence environment. This research contributes to the advancement of Automated residence by highlighting CatBoost as an effective solution for activity recognition. The findings offer valuable insights into CatBoost’s potential for enhancing activity recognition in real-world Automated residence scenarios, leading to improved personalized and intelligent services.