Deployment

Deployment Model Architecture

Deployment of the model APIs

since we want low-latency prediction we used async service instead of sync service

Here are APIs which can be deployed as part of this assignment

• Prediction API
• Feedback API
• Explanation API
• confidence Interval API

prediction API

• This is designed as a low latency async api
• This API expects x1-x15 feature in a json format and returns a price recommendation

• Response shown in the below figure

Feedback API (NotImplemented)

• Where the feedback about the REcommended Price can be given by the user

Explanation (NotImplemented)

• Explanation for the every prediction why the model recommended particular value as prediction

Confidence Interval (NotImplemented)

• Instead predicting single price , if give exact price and range of the price it will be helpful to the user