Crypto ML • Case Study
CoinPulse

AI-powered Crypto Predictor

A cryptocurrency price prediction platform that uses machine learning to forecast market trends. Users can explore real-time crypto data, view price predictions powered by a RandomForest model, and analyze market indicators through an intuitive web interface.

Spline Scene
CoinPulse mobile preview
Scroll to explore

Project Overview

What It Does

Key Metrics

CoinPulse leverages machine learning to analyze cryptocurrency market data and predict whether a coin's price will rise or fall. The system processes data from the top cryptocurrencies, applies sophisticated technical indicators (RSI, MACD, Bollinger Bands), and uses a RandomForest classifier to generate predictions with high accuracy.

Users can select any cryptocurrency from the interface, and the application queries the cloud-hosted ML model via FastAPI to deliver instant predictions based on current market conditions. The React frontend displays real-time data, interactive charts, and trend indicators through an intuitive web interface.

0%

Prediction Accuracy

0%

Uptime

0ms

API Response Time

0+

Supported Coins

0K+

Training Data Points

0+

Technical Indicators

Key Features

Technical Indicators

RSI, MACD, Bollinger Bands, Moving Averages for comprehensive market analysis

Random Forest Classification

Ensemble learning model predicting price movements with high accuracy

Real-time Data

Live crypto data from CoinGecko API for top 100 cryptocurrencies

Self-hosted API

Scalable Flask API deployed on self-hosted VPS with Docker for instant predictions

Project Gallery

Technical Stack

MACHINE LEARNING

• Python
• Scikit-learn
• Pandas
• NumPy

BACKEND

• Flask
• FastAPI
• Gunicorn
• Docker

FRONTEND

• React
• JavaScript
• HTML/CSS

APIS

• CoinGecko API
• REST API

LIBRARIES

• TA-Lib
• SMOTE
• Matplotlib
• Seaborn

Hurdles & Resolutions

Challenges

Challenge

Feature Engineering

Engineering indicator features cleanly for ML training/inference

Challenge

FastAPI Deployment

Packaging and deploying an efficient FastAPI service on Heroku

Challenge

Frontend Integration

Wiring a responsive frontend to live model outputs

Solutions

Solution

Feature Engineering

Created modular pipeline for calculating RSI, MACD, and Bollinger Bands

Solution

FastAPI Deployment

Containerized the application with Docker for consistent deployment

Solution

Frontend Integration

Built RESTful API endpoints with real-time data synchronization