📈 Cryptocurrency Anomaly Detection

This project applies unsupervised machine learning models to detect unusual activities in cryptocurrency trading data. We use Isolation Forest and DBSCAN to identify anomalies in price, trading volume, and market cap that may indicate market manipulations or extreme fluctuations.

---

Problem Statement

Cryptocurrency markets experience frequent fluctuations, but some anomalies might indicate unusual activities such as pump-and-dump schemes, liquidity issues, or significant shifts in investor behavior.
This project aims to detect such anomalies using machine learning, allowing traders and analysts to identify irregular patterns.

---

Approach

1. Exploratory Data Analysis (EDA)

   • Visualize distributions of price, volume, and market cap
   • Detect missing values and perform data cleaning
   • Analyze correlations between different financial metrics

2. Feature Preprocessing

   • Select key numerical features (Price, Market Cap, Volume, Percentage Change)
   • Apply Min-Max Scaling to normalize data

3. Anomaly Detection Models

   • Isolation Forest: Detects outliers based on how easily they can be isolated
   • DBSCAN: Clusters normal data points and labels low-density points as anomalies

4. Results & Analysis

   • Compare anomalies detected by both models
   • Export the results to crypto_anomalies.csv

---

Tools & Technologies

• Programming Language: Python
• Libraries: Pandas, NumPy, Matplotlib, Seaborn, Scikit-learn
• Anomaly Detection Models:
  • IsolationForest (from Scikit-learn)
  • DBSCAN (from Scikit-learn)

---

Exploratory Data Analysis (EDA)

Data Visualization

• Price & Market Cap Distribution
• Trading volume trends over time
• Heatmaps for feature correlation

Sample Distribution Plot

---

1Install Dependencies

pip install pandas numpy matplotlib seaborn scikit-learn