Short-term electricity price prediction optimization method based on similar day-mRMR-regression

The invention relates to the technical field of electricity market transaction, in particular to a short-term electricity price prediction optimization method based on similar day-mRMR-regression, and the method comprises the steps: selecting a similar day sequence set; utilizing mRMR cycle solution to obtain an optimal feature set, and utilizing the optimal feature set to optimize the similar day sequence set again; establishing a regression model based on the optimal feature set of the similar day sequence set, adjusting the number of the optimal features, performing cyclic solution, and performing evaluation by using an average absolute percentage error index to obtain a final regression model; and predicting the short-term electricity price by adopting the final regression model. The problems that an existing prediction method is difficult in electricity price prediction and low in accuracy are solved, a complex function relation determining the electricity price is simplified into a multivariate linear f