Titanic Survivor Prediction - Cartesi Coprocessor

This repo contains an end-to-end template demonstrating a Titanic survivor prediction model running on Cartesi Co-processor's stack. The backend of the Coprocessor dApp is implemented in Python, utilizing m2cgen to generate a machine-learning model for predicting passenger survival based on historical data.

The UI template is implemented with NextJS and Rainbowkit with Viem.

How does it work?

A user can connect their wallet to the frontend and submit passenger details for survival prediction. The frontend sends a transaction to the SurvivorCaller contract, which then interacts with the co-processor to evaluate the passenger's survival probability. The co-processor processes the request using the deployed Python model and returns the result to the on-chain contract, which is then displayed on the UI.

A data flow diagram is provided below to illustrate the flow of data between the frontend, the on-chain contract, and the co-processor.

Project Structure

• backend-cartesi-survivor-py/ - Backend implementation in Python, using m2cgen for model inference.
• contracts/ - Smart contract (SurvivorCaller) handling requests to the co-processor.
• ui-coprocessor-template/ - Frontend React application for user interaction and prediction results.

Setup Instructions for Devnet

1. Run Cartesi-Coprocessor Devnet Environment

Before running the dApp, you need to have the Coprocessor devnet environment running. It will spin up a local operator in devnet mode that will host the dApp backend.

Refer to mugen-docs to understand the components and setup the environment

cartesi-coprocessor start-devnet

To stop and clean up the environment later, use: cartesi-coprocessor stop-devnet

2. Build and Deploy Backend Cartesi Machine

Navigate to backend-cartesi-survivor-py/ and follow the steps in the dedicated README file to generate the model and deploy the backend.

cartesi-coprocessor publish --network devnet

Get the task issuer and machine hash

cartesi-coprocessor address-book

3. Deploy SurvivorCaller Smart Contract

To deploy the contract, cd into the contracts/src folder and run the following command: cartesi-coprocessor deploy --contract-name SurvivorCaller --network devnet --constructor-args <task_issuer> <machine_hash>

4. Run Frontend Application

Navigate to the frontend survivor-ui folder

npm install
nom run dev

The frontend will be available at http://localhost:3000