Hands-On Data Analysis with Pandas

ARIMA

As we discussed in Chapter 1, Introduction to Data Analysis, ARIMA models have autoregressive, difference, and moving average components. This can also be achieved using the statsmodels package. The StockModeler.arima() method returns a fitted ARIMA model for the stock according to the specifications provided:

>>> arima_model = StockModeler.arima(nflx, ar=10, i=1, ma=5)

We're picking these values because they run in a reasonable amount of time. In practice, we can use the autocorrelation_plot() function from pandas.plotting that was introduced in Chapter 5, Visualizing Data with Pandas and Matplotlib, to help find a good value for ar.

Once the model is fitted, information on it can be obtained with the model's summary() method:

>>> ...