Fund Investment Decision in Support Vector Classification Based on Information Entropy
The whole paper is here.
In this project, the support vector classification based on information entropy (IE-SVC) is put forward to improve the accuracy in the field of capital investment decisions. Two classic methods, the K-Nearest Neighbors algorithm (K-NN) and the Radius Basis Function Neural Network (RBFNN), are applied to compare the performance. In the experiment of Gates foundation investment decision, its results show that the IE-SVC can be faster and higher accuracy than those of other methods.
We propose using IE-SVC to improve the accuracy in the field of capital investment decisions. At first, we use SVC to deal with this problem, but the number of negative and postive is not balance. So that, synthetic minority over-sampling technique is used in order to balance the training set. Then, we find that SVC is not a high-performace method in searching the support vector. IE-SVC is chosen to deal with this problem in order to find the support vector fastly.
And we compare other methods (K-NN, RBFNN), and the result is shown as follows:
It is a long day for me to rethink the project I did during my bachelor period. In short, I think this work is a good practise for undergraduate student to experience the research progress. I am super lucky for having a chance to experience the progress comparing my peers in my university, especially in math, a science not engineering, college.