This repository hosts a comprehensive project aimed at time-series prediction using neural networks and the application of three distinct anomaly detection approaches across various datasets.
The primary objectives of this project include:
- Time-Series Prediction: Developing proficiency in time-series prediction through the implementation of neural networks.
- Anomaly Detection: Implementing anomaly detection methods with the application of different datasets.
It intends to implement time-series prediction with neural networks and three approaches(decomposition, prediction, and clustering) for anomaly detection with different data sets. The time-series prediction with neural networks is the base to conduct the prediction-based anomaly detection in this project.
For each anomaly detection task, it includes a chain of sub-tasks from exploratory data analysis, feature extraction, model building, prediction towards anomaly detection.
The project encompasses various tasks structured as follows:
- Exploratory Data Analysis (EDA): The project kicks off with in-depth exploratory data analysis to gain insights into the datasets.
- Feature Extraction: Following EDA, the project focuses on feature extraction, extracting valuable features to support prediction and anomaly detection.
- Model Building: The core phase of the project involves constructing models for time-series prediction and anomaly detection.
- Prediction: Applying the developed models to make predictions based on the data.
- Anomaly Detection: Utilizing the predictive models for the purpose of anomaly detection within the datasets.
This project makes use of various Python libraries to facilitate data analysis, machine learning, and deep learning tasks. The key libraries include:
- Pandas
- Statsmodels
- Scikit-learn
- Keras and TensorFlow
- Additional Common Libraries: Complementary libraries such as NumPy, Matplotlib, and more are employed for various data processing and visualization tasks.
Please refer to the project documentation and code for detailed information on methodologies, datasets, and implementation details.