Anomaly Detection for Time Series Data using Python

This project is a streamlit application that detects anomalies in time series data using the ADTK (Anomaly Detection Toolkit) library in Python. Users can choose from different anomaly detection methods, including ThresholdAD, QuantileAD, and InterQuartileRangeAD (IQR), etc to analyze the dataset and visualize the detected anomalies.

Click Here to View Project

TO-DO

• (feature) Add an option such that the user can upload data via CSV file.
• (feature) Add an option such that the user can tweak/adjust different parameters.

Purpose

The purpose of this small project is to get myself started with the basics of Time Series, anomaly detection, and different techniques/functions available to effectively analyze the data.

My learnings from this project

1. Anomaly Detection Techniques: Gained insights into different anomaly detection techniques such as ThresholdAD, QuantileAD, and InterQuartileRangeAD (IQR) using the ADTK library.
2. Data Preprocessing: Learned how to preprocess and validate time series data for anomaly detection, including handling datetime formats and setting the index appropriately.
3. ADTK Library Usage: Acquired knowledge on how to use the ADTK library for anomaly detection, including how to fit detectors and interpret the results.
4. Parameter Tuning: Experienced the importance of parameter tuning in anomaly detection models to improve the accuracy and relevance of detected anomalies.