https://github.com/yaswanth1702/anomaly-detection-dashboard

https://github.com/yaswanth1702/anomaly-detection-dashboard

business-intelligence data-analytics powerbi sql

Last synced: 4 months ago
JSON representation

• Host: GitHub
• URL: https://github.com/yaswanth1702/anomaly-detection-dashboard
• Owner: Yaswanth1702
• License: mit
• Created: 2025-05-16T22:34:13.000Z (about 1 year ago)
• Default Branch: main
• Last Pushed: 2025-05-16T22:51:10.000Z (about 1 year ago)
• Last Synced: 2025-06-01T22:49:57.656Z (about 1 year ago)
• Topics: business-intelligence, data-analytics, powerbi, sql
• Language: TSQL
• Homepage:
• Size: 2.19 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

          
# Anomaly Detection Dashboard

The **Anomaly Detection Dashboard** is an interactive web application built with Plotly Dash to enable real-time anomaly detection on time-series or tabular datasets. This project provides users with the ability to upload their own CSV data, preprocess it automatically, apply multiple machine learning-based anomaly detection algorithms, and visualize the detected anomalies through dynamic, intuitive graphs.

## Features Implemented

* **Dashboard Interface with Plotly Dash:**

  Developed a responsive and user-friendly dashboard UI for data upload, algorithm selection, and visualization.

* **Multiple Anomaly Detection Models:**

  Implemented three popular unsupervised anomaly detection algorithms from Scikit-learn:

  * Isolation Forest

  * One-Class SVM

  * Local Outlier Factor

* **Interactive Visualization:**

  Created line and scatter plots that dynamically highlight detected anomalies for easy interpretation.

* **Data Upload and Preprocessing:**

  Allowed users to upload custom CSV files. Built automated preprocessing including handling missing values and feature scaling to prepare the data for modeling.

* **Result Export:**

  Added functionality to download anomaly detection results for further offline analysis.

## Technology Stack Used

* **Python** for backend logic and machine learning.

* **Plotly Dash** for building the interactive dashboard UI.

* **Scikit-learn** for implementing anomaly detection algorithms.

* **Pandas & NumPy** for data handling and preprocessing.

* **Joblib** for model serialization and efficient loading.

## Usage Instructions

* Upload your time-series or tabular CSV dataset via the dashboard interface.

* Choose one of the implemented anomaly detection algorithms.

* View the interactive plots highlighting detected anomalies.

* Download the processed results if needed.

## Data Handling

* Supported CSV data with time-series or tabular format.

* Automatically preprocesses data by imputing missing values and scaling features to optimize model performance.

## Future Work (Planned Enhancements)

* Integrate performance metrics like precision, recall, and F1-score.

* Add time-series anomaly trend visualizations.

* Support real-time streaming data.

* Implement alerting and notification features.

## Project Structure (Key Components)

* `app.py`: Main dashboard application script built with Dash.

* `detect_anomalies.py`: Implementation of anomaly detection algorithms.

* `preprocess.py`: Data preprocessing functions including missing data handling and scaling.

* `requirements.txt`: List of dependencies.

* Sample datasets and models organized under dedicated folders.