A recipe web scraper/parser and meal planner designed to optimize nutrition, minimize food waste, and maximize flavor.
Author: Sarah Birmingham
This project was inspired by trying to plan Thanksgiving dinner, and more specifically inspired by the single three-year old container of French's fried onions that I can't seem to use up. When finished, pumpkinpy will be able to amass thousands of recipes from different sources, parse them into usable data, and create potential menus that feature cohesive and complimentary flavors by optimizing the number of ingredients shared between recipes. On a larger scale, pumpkinpy will also create customized meal plans tailored to the user's dietary needs, nutritional requirements, and tastes. As food costs rise nationwide, shared-ingredient-optimization will also help keep grocery bills low and reduce the food waste that inevitably follows inefficient meal planning. Importantly, users will be able to elect to use either a range or a minimum for their nutritional preferences and opt to hide nutritional info in the final output.
The first use case extracts recipes from MinimalistBaker.
Web Scraping
X Extracts all recipes from one website
Extracts recipes from multiple websites
Extracts recipes from any given website
Recipe Parsing
X Reliably parses recipe title and URL where available
X Reliably parses ingredient list into amount/unit/item/notes, where available
X Reliably parses recipe tags/keywords/categories where available
X Reliably parses nutrition info into calories/protein/carbs/fat, where available
Nutrition Inferral
Converts ingredient units into a common factor
Uses systems of equations to determine nutrition content of individual ingredients
Uses open FDA data to determine/confirm nutrition content of individual ingredients
Calculates recipe nutrition info where it's missing
Recipe Tagging
Uses supervised learning techniques to tag untagged recipes based on shared ingredients with
tagged recipes into categories like 'breakfast', 'lunch', 'snack'
Tags untagged recipes as vegetarian, vegan, dairy-free, etc.
Meal Prep Planning
Accepts specified number of meal options as a parameter for a given period
(2 breakfast options, 3 lunch options, 5 days)
Accepts target calories and macronutrients per day as a parameter
Accepts an ED-friendly minimum nutrition requirement as the only parameter
Accepts desired cuisines or ingredients as a parameter
Accepts dietary requirements as a parameter
Creates groups of different recipes that satisfy the above parameters
Rank meal plans by how closely they match input parameters
Rank meal plans by number of unique ingredients (food waste/cost reduction)
Option to hide nutrition info in final output
Menu Planning
Accepts number of dishes as a parameter for a single meal (3 entrees, 6 sides, 2 drinks)
Accepts desired cuisines or ingredients as a parameter
Accepts dietary requirements as a parameter
Creates sample menus that satisfy the above parameters
Rank menus by how closely they match input parameters
Rank menus by number of unique ingredients (food waste/cost reduction, as well as meal cohesion!)
Bonus
Identify when a recipe can be modified to fit certain dietary requirements
Retain info on previously-used recipes to ensure variety
Re-do regex-heavy ingredient parsing in favor of NLP techniques
like the New York Times