This article comprehensively studies modern models and algorithms for intelligent analysis of data from IoT (Internet of Things) sensors. The research used data from various types of sensors in industrial, transport, smart home and bio-health monitoring systems. Large-volume, various-format and noisy data from sensors were first processed through pre-processing stages (Data Cleaning, Normalization, Feature Extraction, Outlier Detection and Data Augmentation). Machine Learning algorithms (KNN, Decision Tree, Random Forest, SVM) and Deep Learning approaches (CNN, RNN, LSTM, Autoencoder) were used to analyze the data. As a result of the experiments, the accuracy and stability indicators were evaluated. The results showed that the LSTM and CNN algorithms provide the highest accuracy (95%) in analyzing visual and timeseries sensor data, while Random Forest provides stable results for noisy and small-volume data. The article also includes recommendations regarding the efficiency of algorithms, the risk of overfitting, the complexity of parameter tuning, and the type of data. The research results have practical implications for forecasting, decision-making, and system efficiency improvement of IoT sensor data in smart transportation, Industry 4.0, smart home, and healthcare systems.