Learning Path Planning Algorithm Based On Multi-dimensional Time Series Data Analysis

With the continuous development of online education,improving the quality of online teaching is an effective means of high-quality development in the education industry.Based on the multi-dimensional time series data generated by learners in the network learning platform,this thesis conducts collaborative analysis on these data,excavates the learning mode of learners,studies the learning path planning algorithm,and recommends learning paths that conform to learners' learning habits and cognitive levels.Online education is used to consolidate offline teaching and improve teaching quality.This thesis proposes two learning path planning algorithms,which are mainly researched on two aspects.The contents are as follows:1.Aiming at the demand that online learners want to master more knowledge with the optimal learning path and the least learning time,this thesis proposes a learning path planning algorithm based on multi-dimensional time series data and ant colony optimization.Firstly,based on the differences in learners' cognitive levels,learners are divided into three levels: Primary,Intermediate and Advanced.Secondly,this algorithm performs collaborative analysis on the multi-dimensional time series data generated by learners,innovatively proposes a method to describe the conceptual interaction degree of the knowledge points,and redefines the initial pheromone and heuristic information of ant colony optimization algorithm,so as to generate learning paths that meet the learners' needs.Finally,the individual learning effects of learners and the teaching effects of learners of different levels are modeled and experimentally compared,dynamically update the teaching duration of knowledge points and allocate suitable learning rhythm for learners of different levels.The effectiveness of the proposed algorithm is verified by using the degree of achievement of course goals.2.Aiming at the problems that the current learning path planning algorithms do not consider learners' learning habits and the conceptual interaction degree of learners to knowledge points,this thesis proposes a learning path planning algorithm based on learning path differences and ant colony optimization.The algorithm is based on the multi-dimensional time series data generated by learners in the network learning platform.Firstly,the attenuation coefficient is introduced to improve the dynamic time warping algorithm,which is used to characterize the learning path differences of the prior learners,and combined with the spectral clustering algorithm to classify the prior learners.Secondly,analyze the learning paths of different types of learners,and combine ant colony optimization algorithm to plan the learning paths.Finally,classify learners according to their actual learning paths,and recommend learning paths that meet the learning habits of different types of learners.The advantages of the algorithm are verified from two aspects:the evaluation index of the clustering algorithm and the degree of achievement of course goals.Using the multi-dimensional time series data generated by learners in the online learning platform to conduct an empirical analysis on the algorithm proposed in this thesis.The experimental results show that the optimization of learning path and teaching duration based on the prior learners' learning situation,it is conducive to improving the degree of achievement of course goals and course completion rate.Experiments verify the effectiveness of the two algorithms proposed in this thesis.