This project aims to develop a cutting-edge anomaly detection system to identify irregularities in industrial IoT data in office settings, potentially preventing equipment failures and optimizing official processes.
Anomaly detection has been used in various data mining applications to find the anomalous activities present in the available data.
In real time deployment , more number of days pass by , hence gathering more training data , accuracy drastically improves.
Real office data from IoT sensors is collected and stored in a CSV file based on real time data, which serves as the basis for model development and predictions.
-
Python 3.9+
-
Jupiter Notebook
Deep Learning frameworks required:
🔢 Numpy 📝 Pandas 🧠 Keras 📈 Matplotlib ⚙️ Scikit-learn
In this project, we have used the Numenta Anomaly Benchmark (NAB) data set. It is a novel benchmark for evaluating ML algorithms in anomaly detection in streaming, online applications.
This dataset also comprises a time-series data named ‘ambient_temperature_system_failure’ in the CSV format. It comprises temperature sensor data of an office setting. This is data for which we know the anomaly causes; no hand labeling.
You can run the 'anomaly_detection.ipynb' file in Jupyter Notebook/Google Colab from the .zip file.
- Ankur Chanda
- Subhadeep Chakraborty
- Soumili Chakraborty
- Diptendu Majumdar