Stock price prediction has always been the focus of investors' attention in the stock market. In recent years, deep learning technology has been widely used in this field. In the era of big data, feature selection is a necessary part of data preprocessing. Feature selection is a data dimensionality reduction technology, and its main purpose is to select the relevant features that are most beneficial to the algorithm from the original data, reduce the dimensionality of the data and the difficulty of learning tasks, and improve the efficiency of the model. This paper has performed analysis of input feature selection with three feature selection methods: Multiple linear regression analysis, Correlation matrix heatmap, Feature importance. Plus the original features set, four different input features sets were provided for predicting stock price of ten China's new energy leading stocks with LSTM. From the conducted experiments, it is found that after using the feature selection method, the prediction results of all ten stocks are performed better than the prediction results under the original features.