After using our models to make a prediction based on test dataset, we plot the comparison between predicted value and true value.

 

Figure 1 shows the comparison between predicting value and true value by using ridge regression. The mean squared error of ridge regression is around 0.0126847. The ridge regression model can predict the “close price” accurately.

 

Figure 2 demonstrate the learning curve for the neural network. And Figure 3 shows the comparison between predicting value and true value by using neural network. The mean squared error of neural network is about 0.54213166. The learning curve illustrates that the model is unstable during epochs. And there might be overfitting situation in the training model. The neural network model does not predict close price as precisely as ridge regression does.