Heat pumps, being complex systems, are susceptible to various malfunctions. By harnessing contemporary IoT technologies, these devices continuously transmit data which enables monitoring, maintenance, and efficiency. This study focuses on identifying compressor short duration cycles as faults through supervised machine learning algorithms such as XGBoost, Random Forest, SVM, and k-NN. Data preprocessing and labeling were conducted using extensive logged data from heat pump systems, addressing issues like high dimensionality, data sparsity, and temporal dependencies. The methodology included feature engineering, interpolation of missing data, and downsampling for compressor short duration cycles. Supervised machine learning models were applied to classify these short duration cycles. Among the models, XGBoost achieved the highest accuracy and F1-scores, effectively distinguishing between normal and fault conditions. The findings highlight the potential of machine learning to enhance predictive maintenance and operational efficiency in heat pumps.