特征分解型求取离散分数阶傅里叶变换 特征分解算法通过求ＤＦＴ的核矩阵Ｆ（ｋ，ｎ）的特征值 和特征向量构造ＤＦＴ核矩阵的分数幂，以此作为Ｆａ（ｋ，ｎ） 来计算ＤＦｒＦＴ。(The kernel matrix eigenvalue decomposition algorithm for DFT F (k, n) characteristic value And the feature vectors are constructed as the fractional powers of the DF

These files are matlab source code for price forecasting for smart meter hourly data. Step 1 relevant features are selected by Gray Correlation, Random Forest, Relief F algorithms. Then Kernel PCA of features are calculated. Price is predicted by Ker