Development of feature engineering tools for multivariate time series
Time series analysis is the key task in several domains such as health diagnosis (for example, electroencephalogram (EEG) analysis), anomaly detection of jet engines, stock market price prediction. In particular, time series data related to several parameters of an instance (for example, a jet engin...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Final Year Project |
| Language: | English |
| Published: |
Nanyang Technological University
2021
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/148839 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Time series analysis is the key task in several domains such as health diagnosis (for example, electroencephalogram (EEG) analysis), anomaly detection of jet engines, stock market price prediction. In particular, time series data related to several parameters of an instance (for example, a jet engine) need to be investigated simultaneously to perform accurate diagnosis or prediction. This project differentiates between correlation and causation as we investigate on the analysis of causality for such multivariate time series datasets. These causality relationships between different time series are established using different tests such as Granger causality and Transfer Entropy. In particular, we look into the implementation of the IDTxl (Information Dynamics Toolkit xl) module for our causality test. To further improve the precision and accuracy of our causality test, topics related to stationarity is explored using the python statsmodels package. Lastly, to promote additional utilization of this research work, an GUI toolkit is implemented in flutter due to its superior widgets catalog. |
|---|