Plot auto_arima predictions on top of actual values

Question

I am still quite new to python and I can't figure out how to do this:

I have a pandas dataframe (data) with two columns: date and values (integers). I feed this dataframe into an auto_arima method.

stepwise_model = auto_arima(data, start_p=1, start_q=1,
                   test='adf',
                 max_p=2, max_q=2, m=7,
                   start_P=0, seasonal=True,
                 d=None, D=1, trace=True,
                   error_action='ignore',  
                   suppress_warnings=True, 
                   stepwise=True)
plt.plot(data,color='r')
plt.plot(stepwise_model.predict(),color='g')
plt.show()

My objective would be to plot the actual values and overlap the values that the model would generate, in order to compare both lines. However, my predict method only gives me ten values.

If I set:

plt.plot(stepwise_model.predict(0,len(data)),color='g')

It gives me an error saying that

raise ValueError('Prediction must have `end` after `start`.')

How can I do this? Maybe there is another method than using predict to plot the correct values?

Answer 1

The ARIMA model here is a different implementation then e.g. statsmodels. The predict() method only takes a single parameter to define the length of the forecast which is by default 10. Instead of:

plt.plot(stepwise_model.predict(0,len(data)),color='g')

You need to give only the amount of time steps you want to predict (no start or end like in statsmodels):

plt.plot(stepwise_model.predict(len(data)),color='g')

This is the model you have: https://alkaline-ml.com/pmdarima/modules/generated/pmdarima.arima.ARIMA.html

And this is it's predict() method: https://alkaline-ml.com/pmdarima/modules/generated/pmdarima.arima.ARIMA.html#pmdarima.arima.ARIMA.predict

If you want to plot the predicted data over the actual data, you can do this here:

plt.plot(stepwise_model.predict(len(data)),color='g')
plt.plot(actual_data)
plt.show()

Everything you put before plt.show() will be in the same figure.