Machine learning algorithm plays a key role in intelligent logistics and equipment support. The basic principles and different characteristics of typical shallow machine learning support vector machine and typical deep learning convolutional neural networks are analyzed. The two kinds of model algorithms are used to model the data of people’s daily activities. Among them, the sample data distinguishes the big sample and the small sample, the four models are used to classify and identify people’s daily activities. The results show, for large sample data, the classification effect of deep learning is obviously better than that of traditional machine learning algorithm, for small sample data, traditional machine learning algorithm is better than deep learning algorithm. The ability of deep learning to deal with big data is obviously better than that of dealing with small sample data. Shallow website support vector machine is better at handling big data than small samples, but not much. The results provide important reference for the choice of algorithm.