Position Prediction Method Based on Improved Decision Tree

Abstract

C4.5 decision tree algorithm is one of the commonly used classification prediction algorithms. It is a tree structure. Its advantage is that the process of obtaining results is easy to understand and the calculation is small, etc. Its disadvantage is that it is easy to cause over-fit, and will be very complex when there are too many categories. In the process of analyzing skill proficiency and positions, different job positions are available for different levels of skill proficiency. When there are more job categories, the prediction effect of decision tree classification is poor and still needs to be improved. In this paper, an improved decision Tree algorithm WF_D-tree is proposed, which adds skill proficiency weight to each data in the data table. Through the calculation method of skill proficiency weight, the relative redundant data of table data are removed. Through the longitudinal calculation method of skill proficiency weight, dimension reduction of data. Experimental results show that the improved decision tree greatly reduces the running time of the decision tree algorithm in the job prediction, and also improves the accuracy of the prediction results. In terms of time, the WF_D-tree is also significantly higher than the decision tree before the improvement. In terms of accuracy, the decision tree was improved by about 11 percentage points.