Neural Network - LSTM Model for Multivariate Time Series Data Prediction

dc.contributor.authorChowdhury, Megdam
dc.contributor.authorShahnewaz, Tahsin
dc.contributor.otherNikolay M. Sirakov
dc.date.accessioned2022-05-03T18:45:18Z
dc.date.available2022-05-03T18:45:18Z
dc.date.issued2022
dc.descriptionTexas A&M University- Commerceen_US
dc.description.abstractTime series forecasting is considered as a dark horse in the field of data science. It is the most critical factor that determines whether the changes in one factor will cause rise or fall in the system environment. Multivariate time series forecasting is an important machine-learning problem where the data involves a mixture of long and short-term pattern, which can selectively record and discard relevant information from the data. Traditional approaches may fail in this process due to time delay and gradient disappearance. In this study, we applied Neural Network- LSTM model to convert a Multivariate time series data into a date time format and used two LSTM layers for sequential analysis of multidimensional data. We validated our neural network using multi attribute stock price historical data for prediction. We used Keras and TensorFlow deep learning library in a Python environment and run our data through the LSTM model to predict the opening prices of 90 days in the future.en_US
dc.identifier.urihttps://hdl.handle.net/11274/13629
dc.language.isoen_USen_US
dc.subjectMathematics
dc.titleNeural Network - LSTM Model for Multivariate Time Series Data Predictionen_US
dc.typePresentationen_US

Files

Original bundle

Now showing 1 - 2 of 2
No Thumbnail Available
Name:
Revised Presentation.mp4
Size:
73.05 MB
Format:
Description:
Presentation video
Loading...
Thumbnail Image
Name:
Chowdhury, M. and Shahnewaz, T. Transcript.pdf
Size:
115.4 KB
Format:
Adobe Portable Document Format
Description:
Presentation transcript

License bundle

Now showing 1 - 1 of 1
No Thumbnail Available
Name:
license.txt
Size:
1.68 KB
Format:
Item-specific license agreed upon to submission
Description: