Extract Mean Absolute Percentage Error (MAPE) score from the ARIMA forecasting tool

[SamuelHamilton]

Dear community, I developed a time series forecast using the ARIMA tool. The forecast predicts a development of an employee base (headcount) over a period of time. 

Now, certain performance/accuracy KPIs such as RMSE, MAPE or AIC are available within the ARIMA tool (see below).

I would like to use the MAPE score to inform the users of this forecast about its accuracy. The score is however not readily available in the output of the tool. 

I've tried to read the underlying R code behind the tool macro to figure out how to extract the score using the R tool, but I've had no success. I've managed to extract the AIC score (see attached workflow), but not the MAPE. It seems like the configuration for the KPI is different from AIC and there are two variables that need to be extracted: fit.stat1 and fit.stat2. That is only my assumption though. 

Does anyone know how to extract the MAPE score from the ARIMA model using the R tool? 

I attach a workflow with a sample dataset, where I perform the forecast and extract the AIC score. 

Thanks for any help in advance!

[DataNath · Accepted solution]

Hey @SamuelHamilton, here's a couple of approaches I can think of. If you're open to generating a report for the user, then you can just Render the results of the model:

If you want to keep these within the workflow, I've filtered out the 'In-sample Error' part of the report and then parsed the HTML of the reporting snippet:

[SamuelHamilton]

Hi @DataNath, this is amazing! I am gonna go with the RegEx option, but both would ultimately work. 

Thanks a lot!!