/outliers

Primary LanguagePython

Outlier Detection Techniques

This Streamlit web application offers a comprehensive exploration of outlier detection techniques applied to a dataset containing age data. Below are the key features and functionalities of the application:

Features:

  1. Dashboard: The application includes a sidebar dashboard that provides easy access to different sections of the app.

  2. Normal Distribution & Z-scores:

    • Visualizes the probability density function (PDF) of age data and highlights outliers using Z-scores.
    • Displays a box plot of age data and identifies outlier values.

  3. Isolation Forest:

    • Applies the Isolation Forest algorithm to detect anomalies in the age data.
    • Visualizes outliers using scatter plots and box plots.

  4. Local Outlier Factor (LOF):

    • Utilizes the LOF algorithm to identify local outliers in the age data.
    • Presents outliers visually through scatter plots and box plots.

  5. Percentile-Based Method:

    • Identifies outliers based on their position in a sorted list using top and bottom percentiles.
    • Displays top and bottom percentile outliers in separate tables.

  6. Winsorization:

    • Implements Winsorization to transform the data by limiting extreme values (outliers) to a specified percentile.
    • Presents both original and winsorized age data in a single table.

Libraries Used:

  • Streamlit
  • Pandas
  • Plotly
  • NumPy
  • Scikit-learn

How to Run:

To run the application locally, follow these steps:

  1. Make sure you have Python installed on your system.
  2. Install the required dependencies using pip:
  3. Navigate to the directory containing app.py in your terminal.
  4. Run the following command:
  5. The application will open in your default web browser, allowing you to explore and experiment with different outlier detection techniques.

Graphs Used:

  • Probability Density Function (PDF) plot
  • Box plot
  • Scatter plot
  • Table (for displaying outlier values)

Contributions and Feedback:

Contributions to the project are welcome! If you encounter any issues or have suggestions for improvement, please feel free to open an issue or submit a pull request on GitHub.

Acknowledgments:

Special thanks to Streamlit for providing an intuitive framework for building interactive web applications with Python.