Estimate Food Delivery System Design
Explore how to design a machine learning system that estimates food delivery times accurately and at scale. Understand data assumptions, system components like feature stores and Kafka, and the user-service interactions. Learn best practices for model training, real-time updates, and system scalability to improve delivery time predictions.
4. Calculation & estimation
Assumptions
For the sake of simplicity, we can make these assumptions:
- There are 2 million monthly active users, a total of 20 million users, 300k restaurants, and 200k drivers deliver food.
- On average, there are 20 million deliveries per year.
Data size
- For 1 month, we collected data on 2 millions deliveries. Each delivery has around 500 bytes related features.
- Total size: 500 * 2 *