TLDR: Given how important the global supply chain is right now, how might we help retailers and CPG companies to keep food on our shelves? Demand forecasting is back in the spotlight these days as Retail and CPG companies struggle to keep up with run-away demand for certain products in our new WFH normal. I'm sure many of you are also being asked about this problem, and I'm wondering how you all are forecasting demand in this environment. COVID-19 qualifies as a'black swan' event: the distribution of product-store-time sales has fundamentally changed over the past two weeks. Ensemble models (e.g., XGBoost, LightGBM) are likely to underfit if trained on historical data, while time-series models are likely to overfit due to high variance within product segments over the past few weeks Many retailers are seeing massive upswings in demand for specific products, but it's hard to tell whether these trends represent a'new normal' level of demand vs. widespread panic purchases that are unlikely to repeat themselves (e.g., the toilet paper craze) How are you thinking about approaching this problem?

"It was a one-day competition with the aim of building machine learning models and increasing the awareness of students on Artificial Intelligence.

The definition that I've found useful in my own research is that AI is machines imitating humans.

"Leveraging that information with Machine Learning to explore different scenarios and learn from those results could be a game changer in finding a …

Data is the raw material that fuels artificial intelligence and machine learning initiatives, but it actually can't be that raw.

Machine learning and AI can augment retailer intelligence to make the impossible, possible.

Similarly, companies offering travel insurance use machine learning and mathematical models to personalize the customer's buying experience.

I know this is a joke, but are there really people out there using linear regression and calling it "machine learning"? Edit: y'all are right that linear regression is certainly a form of statistical learning. I think the main reason I instinctively think of it as being "not machine learning" is that it has a simple closed-form solution--so the "machine" part of statistical learning with linear regression is unnecessary. The same can't be said for a lot of other algorithms that are more comfortably referred to as "machine learning", you know?