The document discusses using additive regression models to predict wholesale produce prices based on long-term variations, seasonal components, and noise. This could help restaurants improve menu pricing and design by predicting food costs. A case study shows how predicting price increases for ingredients like Brussels sprouts could help a catering company maintain profits despite rising costs. Next steps discussed include incorporating origin data and using techniques like Fourier decomposition and feature engineering to improve predictions.
2. Turnover in the $900 billion restaurant industry:
80% of new restaurants fail within 5 years
Sources: https://www.cnbc.com/2016/01/20/heres-the-real-reason-why-most-restaurants-fail.html
https://www.nytimes.com/2016/10/26/dining/restaurant-economics-new-york.html
https://towardsdatascience.com/using-yelp-data-to-predict-restaurant-closure-8aafa4f72ad6
3. Turnover in the $900 billion restaurant industry:
80% of new restaurants fail within 5 years
Poor pricing is a leading cause of failure
- 30-40% of restaurant costs variable
- Typical profit margin just 5%
Sources: https://www.cnbc.com/2016/01/20/heres-the-real-reason-why-most-restaurants-fail.html
https://www.nytimes.com/2016/10/26/dining/restaurant-economics-new-york.html
https://towardsdatascience.com/using-yelp-data-to-predict-restaurant-closure-8aafa4f72ad6
4. Turnover in the $900 billion restaurant industry:
80% of new restaurants fail within 5 years
Poor pricing is a leading cause of failure
- 30-40% of restaurant costs variable
- Typical profit margin just 5%
If food costs can be predicted, menu
pricing/design can be improved
Sources: https://www.cnbc.com/2016/01/20/heres-the-real-reason-why-most-restaurants-fail.html
https://www.nytimes.com/2016/10/26/dining/restaurant-economics-new-york.html
https://towardsdatascience.com/using-yelp-data-to-predict-restaurant-closure-8aafa4f72ad6
16. Case Study: Ora Caters
- Daily meals for a large tech company
- Price negotiated in advance
- Challenge: menu planning at fixed cost
March 2017
Actual
Brussels
Sprouts
30.00
Cabbage 8.10
Carrots 10.50
Apples 10.30
Total $58.90
Recipe:
Shaved
Brussels
Sprouts
Salad
17. Case Study: Ora Caters
- Daily meals for a large tech company
- Price negotiated in advance
- Challenge: menu planning at fixed cost
March 2017
Actual
May 2017
Actual
Brussels
Sprouts
30.00 46.50 (+55%)
Cabbage 8.10 9.60 (+18%)
Carrots 10.50 10.50 (+0%)
Apples 10.30 11.20 (+10%)
Total $58.90 $77.80 (+32%)
Recipe:
Shaved
Brussels
Sprouts
Salad
18. Case Study: Ora Caters
- Daily meals for a large tech company
- Price negotiated in advance
- Challenge: menu planning at fixed cost
March 2017
Actual
May 2017
Actual
May 2017
Predicted
Brussels
Sprouts
30.00 46.50 (+55%) 49.50
Cabbage 8.10 9.60 (+18%) 8.10
Carrots 10.50 10.50 (+0%) 11.10
Apples 10.30 11.20 (+10%) 9.90
Total $58.90 $77.80 (+32%) $78.60
Recipe:
Shaved
Brussels
Sprouts
Salad
21. Next Steps: Origin Information
Typical Origins:
California (Apr-Oct)
Arizona, Mexico (Nov-Mar)
CA Drought
Bumper Crop
Sources:
http://www.businessinsider.com/usda-q3-produce-prices-2012-10
https://www.agweb.com/article/lemon-lovers-get-bitter-shock-on-california-drought/
22. Next Steps: Origin Information
Typical Origins:
California (Apr-Oct)
Arizona, Mexico (Nov-Mar)
CA Drought
Bumper Crop
Sources:
http://www.businessinsider.com/usda-q3-produce-prices-2012-10
https://www.agweb.com/article/lemon-lovers-get-bitter-shock-on-california-drought/
Origin dependent on city, item, time of year!
24. Feature Engineering
Lagging indicators: rolling
averages,historical prices, etc.
Correlated data: similar items,
macro indicators, specialty
producer stock prices, etc.
Slightly better performance
(<5%) than additive
regression model, but
difficult to interpret