فهرست نشریات علمی دسترسی آزاد ایران

In this paper, we define the possibilistic mean, variance and covariance with different risk attitudes and analyses their mathematical property. The mean and variance of the portfolio model are calculated by the fuzzy numbers, which are predicted by the Long-Short Term Memory networks (LSTM), Convolutional Neural Networks (CNN), Support Vector Regression (SVR) and Random Forest (RF). Considering the borrowing constraints, transaction costs and threshold constraints, a new mean and variance fuzzy portfolio selection model with different risk attitudes based on Machine Learning is proposed. Based on the possibilistic measure, the proposed model is transformed into a quadratic programming problem, which is solved by the pivoting algorithm. Finally, the in-sample and out-of-sample comparison analyses of different constraints and different risk attitudes are provided to test the model and the algorithm.