Research On Patent Quality Classification Prediction Of New Energy Vehicles Based On Machine Learning Model

Developing new energy vehicles(NEVs)is the only way for China to move from a major automobile country to a powerful automobile country,and it is a strategic measure to deal with climate change and promote green development.Improving core technology innovation capability is an important support for the high-quality development of the NEVs industry.China’s new energy vehicle market ranks first in the world,but its innovation capacity needs to be improved and its core technologies are not strong.As an important carrier of technical capability,the quality of our new energy vehicle patents is not high,there are problems of "large but not strong,many but not worrying",which hinder the high-quality development of the new energy vehicle industry.The premise of improving patent quality is to identify patent quality efficiently,so it is of great significance to analyze and classify patent quality.Firstly,This paper reviews the development status of China’s new energy vehicle market and new energy vehicle patent.based on existing researches,this paper constructs an evaluation system of patent quality from three dimensions: economic quality,legal quality and technical quality,and uses entropy method to screen out five key indicators from nine influencing factors.Finally,five indexes,including the number of citations,the number of clusters,the number of inventors,the number of IPC and the number of transfers,were selected as the input indexes of cluster analysis and machine learning classification prediction.Secondly,the K-means clustering algorithm was used to perform cluster analysis on 7782 new energy vehicle patents that have been authorized and are in the valid state.The elbow method was used to determine the K value,and then the original patent data was divided into five clustering categories.The results showed that cluster category 4 had the highest patent quality score,cluster category 3 and 2 followed,cluster category 5 ranked the fourth,and cluster category 1 had the lowest patent quality score.Finally,based on the SVM model,BP neural network model and naive Bayes classifier in machine learning,this paper constructed a classification prediction model for the patent quality of new energy vehicles,and measured the performance of the model with four indicators: accuracy,recall rate,accuracy rate and F1 value.The results show that the SVMrbf model with radial basis kernel function has the best classification prediction performance,and the accuracy rate,recall rate,accuracy rate and F1 value of the test set are 97.30%,88.93%,96.94% and 92.76%.The other models are SVMpoly model,BP neural network model and naive Bayes classifier.