PLANIQ - How to import the Metrics like MAPE/MASE etc from the planiq to Anaplan

JagadeshNarayanan

Hi all,

I am working on PlanIQ and managed to get data from the PlanIQ to Anaplan using different algorithms.

now to decide which algorithm is efficient I need the metrics to be imported from the PlanIQ to Anaplan.

So please help me in getting the metrics(MAPE/MASE,etc..).

 

Thanks

 

Jagadesh.S

JagadeshNarayanan

Hi Misbah,

Thanks for your reply.

when I checked a demo from Bedford Consulting in you Tube Link: https://www.youtube.com/watch?v=EyDvzhq4qqg&t=622s

They are showing all the parameters shown in the image below

 

 

 

so they are getting these parameters from the planIQ and based on that they are deciding the best algorithm for every forecast action schedule iteration.

Can you please take a look is it imported manually or  imported directly from PlanIQ 

 

Thanks

Jagadesh.S

timothybrennan

Hi @JagadeshNarayanan ,

 

Importing the metrics that are displayed on the right hand panel of PlanIQ is currently not an available feature.

 

The approach I encourage most users to take is to calculate the relevant forecast evaluation metrics directly in Anaplan.  Once you have forecast results from PlanIQ, if you have actuals available for the period you were forecasting or imported your backtest results, you have all the information you need. 

 

Additionally, the metrics displayed in the PlanIQ interface don't offer much value, as they are simply averages of metrics across the entire data set.  We're working to correct this in the future. By calculating the metrics directly in Anaplan you can correctly determine which particular algorithm has been the most accurate at predicting each individual time series in the data set (which is likely your desired outcome).  

 

I've attached images of a very simple way to calculate the evaluation metrics that I think are usually most appropriate in doing this type of analysis. 

 

 

 

MAE = Mean Absolute Error

MAPE = Mean Absolute Percent Error

sMAPE = Symmetric Mean Absolute Percent Error

RMSE = Root Mean Squared Error

 

Please let me know if there is any other information that would be helpful for you!

 

Thanks for the tag @Misbah 

JagadeshNarayanan

Hi @timothybrennan ,

 

Thanks for your reply. This will solve my problem.

The screen shot you provided for the calculation is very awesome and helpful.

I will try this approach for calculating the different metrics to find the best algorithm.

 

Thanks a lot for your help and information

 

Jagadesh.S