Hands-On Data Analysis with Pandas

Linear regression with statsmodels

Rather than ARIMA, we could build a linear regression of the prior day's closing price to model closing price using statsmodels. This is taken care of by the StockModeler.regression() method:

>>> X, Y, lm = StockModeler.regression(nflx)

Once again, the summary() method can be called to look at statistics on the model's fit:

>>> lm.summary()

This results in the following summary:

The adjusted R² makes this model look very good since it is near one; however, we know that this is simply because stock data is highly autocorrelated, so let's look at the residuals again:

>>> StockModeler.plot_residuals(lm)

This ...