A Gradient Boosting Based Classification Technique for Assisted Prediction Algorithm Research

Classification is a significant data mining technique which offers a wide range of applications to classify the manifold types of data existing in almost every industry of our lives such as predictors, the system of weather forecasting, and business intelligence etc. Both supervised learning and unsupervised learning can be used in this technology, and supervised data mining technology has higher accuracy and performance. As a result, in sensitive applications, monitored data extraction technology is used for forecasting and classification.This paper proposes an auxiliary prediction model based on machine learning through a series of sample feature data. The auxiliary prediction approach makes use of more detailed data, which both increases the accuracy of automatic diagnosis and lowers diagnosis error. The suggested auxiliary diagnosis prediction model will also assist in handling various types of data, which can enhance efficiency and save some time to some level. This study used 918 pieces of relevant data from the UCI machine learning knowledge base. This paper used Decision Tree, Random Forest, Gradient Boosting and Artificial Neural Network algorithm for data sets that were analyzed, and found Gradient Boosting in the data set has better performance, whose accuracy rate has reached 86.67%. The results show that Gradient-boosting-based classification technique can be effectively applied in assisting prediction research.