The invention discloses a big data processing method for a two-stage depth learning model based on sectionalization. The big data processing method based on a sectioned two-stage depth study model comprises steps of 1) constructing and training a two-stage depth study model based on sectionalization, wherein the two-stage depth study model is divided into two stages from vertical levels: a first stage and a second stage, each level of the first stage is divided into M sections transversely, and weight between adjacent neuron nodes in different sections in the first level is 0, 2) dividing big data to be processed into M subsets according to data types, respectively inputting the M subsets into M sections of the first level of the two-stage depth learning model based on sectionalization for processing, and 3) outputting a big data processing result. The two-stage depth learning model based on sectionalization can effectively reduce a model scale and shortens training time of the model, and the big data processing method can improve big data processing speed and shortens processing time.