Machine Learning for Demand Forecasting in Supply Chains
Вставка
- Опубліковано 21 лип 2024
- This webinar will present you the pitfalls and best practices of using machine learning to forecast demand. We'll also see what extra accuracy you can expect from using this technology and how you should plan your first project.
- Наука та технологія
Created one year ago, and is still relevant today! Watched the whole thing and probably will watch it again. Thanks Nicolas! Love this!
any code or tutorial?
I share the code in my books: www.amazon.com/stores/Nicolas-Vandeput/author/B07KL86HMV?ref=sr_ntt_srch_lnk_2
Hi Nicholas, when it comes to feature engineering for future covariates, which features are a must according to you? The only future feature I've been able to implement is the lag features, however one is then constrained by the lowest lag feature, i.e if you have lag 7 day feature, you can't predict further than 7 days into the future. What other future variables are there that one would know in the future, apart from holidays and company specific things like marketing costs, promotions etc?
Great webinar, thanks to share it in UA-cam. Putting a very complex machine learning model in production will not guarantee full automation. You need continuous monitoring to ensure that your model is not drifting, the quality of data is not dropping or the scope of analysis did not change. If you do not have the resources, you'd better rely on a solid simple statistical model or a simple explainable model based on XGBG/LGBM.
XGB/LGBM are the best ;)
What are the best models in your experience?
LightGBM and XGB!
Hi @Nicolas - Do you offer any online training courses on data science for forecasting and effective inventory management?
Yes, sure, you can find them here: supchains.com/live-training/live-training-demand-forecasting/
Thanks for this webinar. You did a great job giving a high level explanation on the ML concept.
I was expecting to have comparaisons between models.
Data Scientists should focus more on showing the results of experimentation then advertising the concept.
I still don’t know of I should invest money and time to build a POC
I discuss results in various case studies. Here are some,
Manufacturer with promotions: 20% forecast improvement nicolas-vandeput.medium.com/forecasting-case-study-ml-driven-forecasts-for-a-manufacturer-with-promotions-3a4dea8a9160
Chemical company: 20% forecast improvement nicolas-vandeput.medium.com/forecasting-case-study-with-a-chemical-company-35d02256667e
Pharma distributor: 25% forecast improvement nicolas-vandeput.medium.com/an-end-to-end-supply-chain-optimization-case-study-part-1-demand-forecasting-2f071b81a490
Retailer with promotions and pricing: 30% forecast improvement nicolas-vandeput.medium.com/using-machine-learning-to-forecast-sales-for-a-retailer-with-prices-promotions-aab9b35d16a
What is missing in your approach I believe : "what is the customer of the forecast I'm building"?
Hello Olivier, not sure I get your comment. Would you mind to re-phrase it?
Data quality and incorporating relevant inputs/predictors based on expert domain knowledge are extremely important. But the dude constantly rubbing his face and fidgeting had me on the edge lol. Which models have you found work best? RandomForest, XgBoost, Facebook Prophet, ESRNN, and DeepAR models all seem promising to me.
It is often more a question of parameter fine-tuning and feature selection than model selection. Nevertheless, I am currently mostly using LGBM/XGBoost.
Can I get the slides?
I (usually) share the slides with all the registered people.
You can be informed of new webinars by joining my mailing list here: mailchi.mp/supchains.com/newsletter