Artificial Intelligence
Predicting Order Wait Time using Machine Learning
In order to provide a better customer experience, one of the world’s largest full-service dining companies required a solution that leverages machine learning to predict how long customers will have to wait for their order to be served.
Client Challenges and Requirements
- Limited factors available in real-time
- Online and table-top orders data segregation
Bitwise Solution
Bitwise created a prediction model to help the client give accurate wait times to customers who have ordered food.
Strategy to distinguish real-time (high frequency) and static (low frequency) factors
Implementation of caching technique for static factors in DynamoDB
Development of statistical standardization methods and advanced feature engineering for online and table-top order segregation
Real-time predictions using end-to-end pipelines on AWS Lambda and API gateway
Tools & Technologies We Used
Python
scikit-learn
AWS Sagemaker
AWS Lambda
AWS API Gateway
AWS DynamoDB
Key Results
Predicted order wait time with 87% accuracy
Improved customer experience
Increased customer base
