Enhancing Time Series Forecasting Accuracy and Resilience in High-Frequency Data Environments through Hybrid Deep Learning Smoothing Models

Lingapandiyan, Dineshkumar (2024) Enhancing Time Series Forecasting Accuracy and Resilience in High-Frequency Data Environments through Hybrid Deep Learning Smoothing Models. Masters thesis, Dublin, National College of Ireland.

Preview	PDF (Master of Science) Download (1MB) | Preview
Preview	PDF (Configuration Manual) Download (913kB) | Preview

Abstract

Flight delays significantly impact airline operations and passenger satisfaction therefore making accurate forecasting is essential for improving scheduling and resource management. However forecasting flight delays presents unique challenges due to the high-frequency noisy and non stationary nature of the data. Traditional time series models such as ARIMA, perform poorly with the non-linear dependencies and sudden fluctuations characteristic of flight delay data. Convolutional Neural Networks (CNNs) and Long Short Term Memory (LSTM) networks on the other hand, are deep learning models that have demonstrated encouraging outcomes when managing intricate temporal patterns However overfitting computing demands and the requirement for big datasets continue to be problems for them. This study proposes a hybrid deep learning smoothing model to address these challenges. By integrating LSTM and CNN architectures with traditional smoothing techniques, such as moving averages and the Kalman filter the hybrid model leverages the strengths of both methods. The deep learning components capture complex temporal dependencies, while the smoothing techniques reduce noise and enhance model stability. Using historical flight data weather information and operational variables the hybrid model demonstrates superior predictive accuracy and resilience compared to traditional methods. Experimental results indicate that the proposed approach outperforms standalone models in terms of mean absolute error and root mean squared error highlighting its robustness in handling high frequency volatile data. This research offers valuable insights for the aviation industry with potential benefits extending to passenger satisfaction operational efficiency, and resource optimization. The hybrid model also has broader applicability in other high frequency data environments such as financial markets and energy management.

Item Type:	Thesis (Masters)
Supervisors:	Name Email Rifai, Hicham UNSPECIFIED
Subjects:	Q Science > QA Mathematics > Electronic computers. Computer science T Technology > T Technology (General) > Information Technology > Electronic computers. Computer science H Social Sciences > HD Industries. Land use. Labor > Specific Industries > Aviation Industry Q Science > Q Science (General) > Self-organizing systems. Conscious automata > Machine learning
Divisions:	School of Computing > Master of Science in Data Analytics
Depositing User:	Ciara O'Brien
Date Deposited:	03 Sep 2025 11:24
Last Modified:	03 Sep 2025 11:24
URI:	https://norma.ncirl.ie/id/eprint/8736

Actions (login required)

View Item