Deep student performance prediction method based on multi-layer LSTM

Abstract

The invention discloses a deep student performance prediction method based on multi-layer LSTM, and the method comprises the following steps: data preprocessing: dividing demographic data according totypes, carrying out the One-Hot coding, obtaining a demographic sequence, carrying out the processing of click stream data according to types, obtaining a click stream sequence, and obtaining preprocessed data after splicing; establishing a multi-layer full connection layer, and inputting a demographic sequence; establishing a multi-layer LSTM structure layer, and inputting a click stream sequence in the preprocessed data; splicing output sequences of the multiple full connection layers and the multiple LSTM structure layers; building a Softmax layer, converting the spliced output sequence into probability distribution, and the category with the large probability is an output result; according to the invention, the multi-layer LSTM structure layer is used as a basic framework, so that theproblems of incomplete data utilization and poor prediction effect are solved; an attention mechanism is introduced, different attention is distributed to different input items, and the prediction effect is improved.

Description

technical field [0001] The invention relates to the research field of natural language processing sentiment analysis, in particular to a deep student performance prediction method based on multi-layer LSTM. Background technique [0002] Thanks to the development of Internet technology, people can easily access information that is several times larger than before, and their horizons and horizons have been greatly expanded. In the Internet age, people can not only listen to the teacher's lectures face to face in the traditional classroom, but also obtain excellent classroom videos and teaching resources from all over the world without leaving home, which undoubtedly provides great opportunities for students' independent learning. However, it also brings many new problems and challenges. Affected by the epidemic this time, schools around the world have changed offline courses to online teaching, and a large number of students have begun to use online education platforms for co...

AI Technical Summary

Problems solved by technology

[0004] The main purpose of the present invention is to overcome the shortcomings and deficiencies of the prior art, provide a deep student performance prediction method based on multi-layer LSTM, and use the LSTM network structure as the basic structure to fundamentally solve the existing problems of the previous student performance prediction models. The problem of incomplete data utilization and poor prediction effect also solves the two problems of cold start and course transfer in the student performance prediction task

Images