Hyperparameter optimisation for ARIMA(p,d,q)

Quant Trading Strategies & Ideas
1

Course Name: Quantitative Trading Strategies and Models, Section No: 3, Unit No: 16, Unit type: Notebook



Shouldn't we do a grid search over cartesian product of the p and q ranges? I'm not sure I understand why we're looking at p, and q individually. 



Many thanks

2

Hi Alex,



In Section 3, Unit 23, the part where we calculate and store the AIC scores for different p,q, and dist values, we use a nested loop for each of those parameters. That way we in effect create a 3-D grid of p * q * dist and store AIC values for all possible combinations. 



Then the minimum AIC score is chosen.



Hope this helps!