Prediction of Urban Innovation Based on Machine Learning Method

Based on regression tree and support vector regression in machine learning, this paper predicts the innovation output of Chinese cities. The results show that the prediction error of regression tree in the test set is 0.28, while the prediction error of support vector regression in the test set is 0.33. Therefore, the prediction effect of regression tree is better than that of support vector regression. In addition, we find that GDP, human capital level and foreign investment all have a positive impact on urban innovation output. The GDP index which represents the economic base has the greatest impact on the innovation output of a city. It means that innovative output depends on a sound economic foundation.