Research On The Influence Of Green Investment On Green Ecology In China

With the reform of market-oriented economic system,China’s futures market is developing gradually.In October 1990,the establishment of Zhengzhou Grain Wholesale Market marked the birth of China’s futures market.After 30 years,China’s futures market has begun to take shape.Futures market is playing an increasingly important role in the process of continuous improvement.The trend of futures prices not only affects the vital interests of investors,but also affects the overall healthy development of the market.Therefore,improving the forecasting accuracy of its not only has theoretical significance,but also has profound practical significance.However,futures prices have the characteristics of nonlinearity and non-stationarity,which brings challenges to high-precision prediction.In order to improve the model prediction effect,increase the robustness on the model in the data set,the author will carry out research from the following contents:(1)Examine the predictability of the market: in this article,through Hurst test,ADF test and variance ratio test proved that the market doesn’t obey random walk hypothesis,has a certain predictability;(2)Collect the variables that are correlated with the prediction variables,use mutual information and Lasso method to conduct comprehensive screening of the collected variables,use stack autoencoder to reduce the dimensionality of the filtered variables,and improve the convergence speed of the subsequent prediction model without losing effective information;(3)Introduce attention mechanism on the LSTM model,and use cuckoo search algorithm to optimize the parameters of the model to improve the robustness and prediction accuracy of the prediction system;(4)In order to reflect the integrity of the prediction system,LSTM model is introduced to predict the uncertainty interval on the quantile regression,and the grid search method is used to determine the optimal parameter combination.The empirical research selects three futures price data.The experimental results show the deterministic predicted MAPE values are below 0.6%,and the interval predicted PICP values are above 90%.Compared with other benchmark models,the prediction system proposed in this study has smaller statistical error.In addition,the prediction analysis of other futures data also verifies this system has good generalization performance.