How to interprete Accumulated Local Effects plot

[xiaochi-liu]

Hello, I'm reading Chapter 8.2 "Accumulated Local Effects (ALE) Plot". May I ask how to interpret the Accumulated Local Effects (ALE) Plot?

In Section 8.2.3 "Estimation", it says:

The value of the ALE can be interpreted as the main effect of the feature at a certain value compared to the average prediction of the data. For example, an ALE estimate of -2 at $x_j=3$ means that when the j-th feature has value 3, then the prediction is lower by 2 compared to the average prediction.

However, in another book talking about ALE, the interpretation is different, and they can even draw partial dependence and accumulated local effect in the same plot, such as Figure 18.5.

I was wondering, is there something different in the theory?

Your kind guidance is much appreciated!

[xiaochi-liu]

I'm sorry for the repeated question but really eager to get the correct understanding of the ALE plot. The book says

ALE plots are centered at zero.This makes their interpretation nice, because the value at each point of the ALE curve is the difference to the mean prediction.

May I ask, what does this mean prediction refer to specifically? Is it just the average of the model's predicted values?

Your kind guidance is much appreciated!

[christophM]

You can plot both curves in the same plot when you add the mean prediction to the ALE curve. Often they also will look similar.

[xiaochi-liu]

Thank you very much for your guidance @christophM! Crystal clear.