College data analysis based on multi-learning method

摘要

This paper proposes a new approach to data analysis based on multi-learning method. The proposed method employs a combination of principal component analysis, K-means clustering, random forest, deep neural network and Jack Philips ROI evaluation model, and is effective in solving the problem of choosing donation recipients. Results are validated through data envelop analysis and are believed to accurately reflect the reality. This research can be helpful to analyses of large scale economic problems.