Download the Excel here and practice with me : abcsupplychain.com/forecast-accuracy-formula-excel/ Join my Free Demand Forecasting WORKSHOP : abcsupplychain.com/demand-forecasting-webinar/?
Hi Edouard. How do you calculate % Accuracy with the MAE method if the actual sales is much lower than the forecast e.g. Forecast = 100, Actual Sales = 10. In that case the % Accuracy would be negative -800%. Do you cap it 0% in that case? How do you deal with negative values for % accuracy? Thanks
Thanks Edouard for your excellent explanation. But for the error calcuation, should it be ( Acual - Forecasted value ??) As you made it Forecasted demand- Actual demand, specailly when it goes to MAPE later it will be affected.
Thank you for the video! Is these topics covered in your inventory management expert course ?If not are you planning for any course related to forecasting and demand planning?
Hi ! Yes, this topic is covered in the Inventory Management Expert course. I'm also launching very soon a course specifically on Forecasting and we will go deeper than in this quick video. More news coming very very soon 😉
Thanks for good recording! Could you explain how to calculate the erros for more than one period of forecast? For example for wk32/wk33/wk34 (forecast updated few times) ... and compare these amounts to received/actual demand?
Hello, do you have excel file for the dashboard you mentioned at the end of video? Is it possible to share that as well? Thanks for the free excel by the way
Thanks Edouard for this great video! You say Demand ≠ Sales and I understand this, but then what is the difference between demand and forecast if Demand ≠ Sales?
2 notes: 1 - The Bias/accuracy illustration is really talking about precision on the x axis. The top right was precise but not accurate, bottom left was not biased or precise but it was accurate (on average). 2 - You are quite dismissive of MAPE but I think this shows your bias towards product supply - where each item has high value variability then MAPE loses relevance, however, in other contexts where value per item is essentially equal (such as call center demand forecasts) then MAPE is standard and with good reason.
I agree with you on your second point, In thie example shown, it doesn't make sense since MAPE is used to calculate an average over multiple periods. However, if you want to calculate one period over different values (families or products), what you can do is weigh each family by its value and then sum the values at the end (abs(error)/demand)*weight. This way, you obtain a more realistic value.
@@drethes yeah. My understanding is that MAPE is a start point that's good for some situations, but you then use one of the variations (such as weighted) to manage the idiosyncrasies of the data.
![Power BI DAX Tutorial - Beginner to Advanced [Full Course]](http://i.ytimg.com/vi/4rC9Ow76n0U/mqdefault.jpg)
Download the Excel here and practice with me : abcsupplychain.com/forecast-accuracy-formula-excel/
Join my Free Demand Forecasting WORKSHOP : abcsupplychain.com/demand-forecasting-webinar/?
Superb: s usual!!!
Great work!
I'm glad you enjoyed it 😊
Hi Edouard. How do you calculate % Accuracy with the MAE method if the actual sales is much lower than the forecast e.g. Forecast = 100, Actual Sales = 10. In that case the % Accuracy would be negative -800%. Do you cap it 0% in that case? How do you deal with negative values for % accuracy? Thanks
I need to understand it to
Thanks Edouard for your excellent explanation. But for the error calcuation, should it be ( Acual - Forecasted value ??)
As you made it Forecasted demand- Actual demand, specailly when it goes to MAPE later it will be affected.
Thank you for your great work and tutorial. have just downloaded the file !!!!
I like the way you explain this complex subject! thanks
Thank you Gabriele ! I like to stay simple and straight to the point 😀
Really very simple and amazing way u explained ❤
Thank you, much appreciated 🙏
It's really helpful
I'm Glad 👍
Thank you for the video! Is these topics covered in your inventory management expert course ?If not are you planning for any course related to forecasting and demand planning?
Hi ! Yes, this topic is covered in the Inventory Management Expert course. I'm also launching very soon a course specifically on Forecasting and we will go deeper than in this quick video. More news coming very very soon 😉
Hi Edourd, need your guidance how to measure forecast accuracy in case of a consumer fragnance businesses to businesses model
Thanks for good recording! Could you explain how to calculate the erros for more than one period of forecast? For example for wk32/wk33/wk34 (forecast updated few times) ... and compare these amounts to received/actual demand?
Hello, do you have excel file for the dashboard you mentioned at the end of video? Is it possible to share that as well? Thanks for the free excel by the way
Thanks Edouard for this great video! You say Demand ≠ Sales and I understand this, but then what is the difference between demand and forecast if Demand ≠ Sales?
How to connect with you on business using your tool
2 notes:
1 - The Bias/accuracy illustration is really talking about precision on the x axis. The top right was precise but not accurate, bottom left was not biased or precise but it was accurate (on average).
2 - You are quite dismissive of MAPE but I think this shows your bias towards product supply - where each item has high value variability then MAPE loses relevance, however, in other contexts where value per item is essentially equal (such as call center demand forecasts) then MAPE is standard and with good reason.
I agree with you on your second point, In thie example shown, it doesn't make sense since MAPE is used to calculate an average over multiple periods. However, if you want to calculate one period over different values (families or products), what you can do is weigh each family by its value and then sum the values at the end (abs(error)/demand)*weight. This way, you obtain a more realistic value.
@@drethes yeah. My understanding is that MAPE is a start point that's good for some situations, but you then use one of the variations (such as weighted) to manage the idiosyncrasies of the data.
What if over forecast. Say actually need 10 pc and I forecast 40 pc. The error is 30. error% is 300%. And accuracy is 200%???
Then accuracy is -200% not 200%. Apply the formula as it is in the file abcsupplychain.com/forecast-accuracy-formula-excel/