Data Science & Machine Learning for Demand Forecasting

In this webinar, I discuss the steps required to build your dream forecasting engine.
- Why do we forecast demand
- Select the right forecasting metric
- Censoring shortages to capture unconstrained demand (and stop to forecast sales!)
- How ML works and why it is so much better than statistical models.
This webinar is based on my latest book, Demand Forecasting Best Practices.
You can download/order it here: www.manning.com/books/demand-...

[@davidtiefenthaler7753]

Would be interesting to get you opinion on MAPE to compare multiple forecasts (or to use as performance metric for to evaluate multiple time series), since RMSE, MAE are not suitable to do so.

[@nicolasvandeput-SupChains]

Hello David, long story short: MAPE is never a good idea.
MAE is fine for comparing different products.

[@davidtiefenthaler7753]

@@nicolasvandeput-SupChains how would you use MAE to compare different product on different scale? Since the MAE does reflect the different scale and is therefore hard to use for comparison or as an aggregated metric of multiple products.

[@davidtiefenthaler7753]

I just noticed your definition of MAE might be different to the standard one (en.wikipedia.org/wiki/Mean_absolute_error) since you represent it as a percentage value. Would be great if you can clarify this.

[@nicolasvandeput-SupChains]

@@davidtiefenthaler7753 MAE scales perfectly if you have many products.
%MAE doesn't scale across different product. it's all explained here:
- www.manning.com/books/demand-forecasting-best-practices
- towardsdatascience.com/forecast-kpi-rmse-mae-mape-bias-cdc5703d242d
In general, no KPIs are perfect. Especially when looking at broad portfolio.

[@muhammadhammadmasood8728]

Awesome session! I'm curious, how would we forecast zeroes? lets say we have inventory for such items but they do no sell at particular time period may be.

[@lextor99]

use static rules based on competitor price

[@neelamyadav533]

Please make video on forecasting intermittent time series data. I tried croston, tsb etc but results are pretty bad.I have only 8 months data . Will you please suggest some methods.

[@nicolasvandeput-SupChains]

With only 8 months, it'll be difficult. But I will make more content on this ;)

[@Terracotta-warriors_Sea]

please make a video on forecasting of slow moving intermittent and lumpy demand patterns such as those encountered in MRO parts demands. How to use Croston method to forecast mean demand and its variance/std dev and then how datascience forecasting can help in such cases.

[@nicolasvandeput-SupChains]

Croston is not a good idea: towardsdatascience.com/croston-forecast-model-for-intermittent-demand-360287a17f5f

[@nwabuezeprecious457]

@@nicolasvandeput-SupChains yes I noticed it did poorly with my dataset. TSB and ADIDA did better.

[@nicolasvandeput-SupChains]

@@nwabuezeprecious457 How do they compare to a moving average 12 months?

[@nwabuezeprecious457]

@@nicolasvandeput-SupChains TSB performed better. but this due to the volatility of my demand data. but if one has a relatively stable demand then a 12 months rolling forecast is suitable. But as you know @Nicolas whatever the technique for forecasting you intend to use will really depend on your case study. Thanks Nicolas for responding reading your book currently.

[@DarkTobias7]

Do you have the github python code available for these?

[@nicolasvandeput-SupChains]

No, but I share them in books available here: www.amazon.com/stores/Nicolas-Vandeput/author/B07KL86HMV?ref=sr_ntt_srch_lnk_2