Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning

Deep learning techniques have significantly advanced time series prediction by effectively modeling temporal dependencies, particularly for datasets with numerous observations. Although larger datasets are generally associated with improved accuracy, the results of this study demonstrate that this a...

Full description

Saved in:
Bibliographic Details
Main Authors: Ana Lazcano, Pablo Hidalgo, Julio E. Sandubete
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Applied Sciences
Subjects:
Online Access:https://www.mdpi.com/2076-3417/14/23/11081
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1846124448073121792
author Ana Lazcano
Pablo Hidalgo
Julio E. Sandubete
author_facet Ana Lazcano
Pablo Hidalgo
Julio E. Sandubete
author_sort Ana Lazcano
collection DOAJ
description Deep learning techniques have significantly advanced time series prediction by effectively modeling temporal dependencies, particularly for datasets with numerous observations. Although larger datasets are generally associated with improved accuracy, the results of this study demonstrate that this assumption does not always hold. By progressively increasing the amount of training data in a controlled experimental setup, the best predictive metrics were achieved in intermediate iterations, with variations of up to 66% in RMSE and 44% in MAPE across different models and datasets. The findings challenge the notion that more data necessarily leads to better generalization, showing that additional observations can sometimes result in diminishing returns or even degradation of predictive metrics. These results emphasize the importance of strategically balancing dataset size and model optimization to achieve robust and efficient performance. Such insights offer valuable guidance for time series forecasting, especially in contexts where computational efficiency and predictive accuracy must be optimized.
format Article
id doaj-art-963f426837ef49558b91e6f19eb4e2d2
institution Kabale University
issn 2076-3417
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Applied Sciences
spelling doaj-art-963f426837ef49558b91e6f19eb4e2d22024-12-13T16:22:45ZengMDPI AGApplied Sciences2076-34172024-11-0114231108110.3390/app142311081Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep LearningAna Lazcano0Pablo Hidalgo1Julio E. Sandubete2Faculty of Law, Business and Government, Universidad Francisco de Vitoria, 28223 Madrid, SpainFaculty of Law, Business and Government, Universidad Francisco de Vitoria, 28223 Madrid, SpainFaculty of Law, Business and Government, Universidad Francisco de Vitoria, 28223 Madrid, SpainDeep learning techniques have significantly advanced time series prediction by effectively modeling temporal dependencies, particularly for datasets with numerous observations. Although larger datasets are generally associated with improved accuracy, the results of this study demonstrate that this assumption does not always hold. By progressively increasing the amount of training data in a controlled experimental setup, the best predictive metrics were achieved in intermediate iterations, with variations of up to 66% in RMSE and 44% in MAPE across different models and datasets. The findings challenge the notion that more data necessarily leads to better generalization, showing that additional observations can sometimes result in diminishing returns or even degradation of predictive metrics. These results emphasize the importance of strategically balancing dataset size and model optimization to achieve robust and efficient performance. Such insights offer valuable guidance for time series forecasting, especially in contexts where computational efficiency and predictive accuracy must be optimized.https://www.mdpi.com/2076-3417/14/23/11081time series forecastingpreprocessingMLPLSTMtransformer
spellingShingle Ana Lazcano
Pablo Hidalgo
Julio E. Sandubete
Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
Applied Sciences
time series forecasting
preprocessing
MLP
LSTM
transformer
title Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
title_full Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
title_fullStr Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
title_full_unstemmed Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
title_short Walking Back the Data Quantity Assumption to Improve Time Series Prediction in Deep Learning
title_sort walking back the data quantity assumption to improve time series prediction in deep learning
topic time series forecasting
preprocessing
MLP
LSTM
transformer
url https://www.mdpi.com/2076-3417/14/23/11081
work_keys_str_mv AT analazcano walkingbackthedataquantityassumptiontoimprovetimeseriespredictionindeeplearning
AT pablohidalgo walkingbackthedataquantityassumptiontoimprovetimeseriespredictionindeeplearning
AT julioesandubete walkingbackthedataquantityassumptiontoimprovetimeseriespredictionindeeplearning