Variable Selection in Time Series Forecasting Using Random Forests

Variable Selection in Time Series Forecasting Using Random Forests

Time series forecasting using machine learning algorithms has gained popularity recently. Random forest is a machine learning algorithm implemented in time series forecasting; however, most of its forecasting properties have remained unexplored. Here we focus on assessing the performance of random f...

Full description

Bibliographic Details
Main Authors: Hristos Tyralis, Georgia Papacharalampous
Format: Article
Language:English
Published: MDPI AG 2017-10-01
Series:Algorithms
Subjects:
ARFIMA
ARMA
machine learning
one-step ahead forecasting
random forests
time series forecasting
variable selection
Online Access:https://www.mdpi.com/1999-4893/10/4/114
id doaj-379545664de44f2485d38e3f9da8bfc3
record_format Article
spelling doaj-379545664de44f2485d38e3f9da8bfc32020-11-24T22:52:54ZengMDPI AGAlgorithms1999-48932017-10-0110411410.3390/a10040114a10040114Variable Selection in Time Series Forecasting Using Random ForestsHristos Tyralis0Georgia Papacharalampous1Department of Water Resources and Environmental Engineering, School of Civil Engineering, National Technical University of Athens, Iroon Polytechniou 5, 157 80 Zografou, GreeceDepartment of Water Resources and Environmental Engineering, School of Civil Engineering, National Technical University of Athens, Iroon Polytechniou 5, 157 80 Zografou, GreeceTime series forecasting using machine learning algorithms has gained popularity recently. Random forest is a machine learning algorithm implemented in time series forecasting; however, most of its forecasting properties have remained unexplored. Here we focus on assessing the performance of random forests in one-step forecasting using two large datasets of short time series with the aim to suggest an optimal set of predictor variables. Furthermore, we compare its performance to benchmarking methods. The first dataset is composed by 16,000 simulated time series from a variety of Autoregressive Fractionally Integrated Moving Average (ARFIMA) models. The second dataset consists of 135 mean annual temperature time series. The highest predictive performance of RF is observed when using a low number of recent lagged predictor variables. This outcome could be useful in relevant future applications, with the prospect to achieve higher predictive accuracy.https://www.mdpi.com/1999-4893/10/4/114ARFIMAARMAmachine learningone-step ahead forecastingrandom foreststime series forecastingvariable selection
collection DOAJ
language English
format Article
sources DOAJ
author Hristos Tyralis
Georgia Papacharalampous
spellingShingle Hristos Tyralis
Georgia Papacharalampous
Variable Selection in Time Series Forecasting Using Random Forests
Algorithms
ARFIMA
ARMA
machine learning
one-step ahead forecasting
random forests
time series forecasting
variable selection
author_facet Hristos Tyralis
Georgia Papacharalampous
author_sort Hristos Tyralis
title Variable Selection in Time Series Forecasting Using Random Forests
title_short Variable Selection in Time Series Forecasting Using Random Forests
title_full Variable Selection in Time Series Forecasting Using Random Forests
title_fullStr Variable Selection in Time Series Forecasting Using Random Forests
title_full_unstemmed Variable Selection in Time Series Forecasting Using Random Forests
title_sort variable selection in time series forecasting using random forests
publisher MDPI AG
series Algorithms
issn 1999-4893
publishDate 2017-10-01
description Time series forecasting using machine learning algorithms has gained popularity recently. Random forest is a machine learning algorithm implemented in time series forecasting; however, most of its forecasting properties have remained unexplored. Here we focus on assessing the performance of random forests in one-step forecasting using two large datasets of short time series with the aim to suggest an optimal set of predictor variables. Furthermore, we compare its performance to benchmarking methods. The first dataset is composed by 16,000 simulated time series from a variety of Autoregressive Fractionally Integrated Moving Average (ARFIMA) models. The second dataset consists of 135 mean annual temperature time series. The highest predictive performance of RF is observed when using a low number of recent lagged predictor variables. This outcome could be useful in relevant future applications, with the prospect to achieve higher predictive accuracy.
topic ARFIMA
ARMA
machine learning
one-step ahead forecasting
random forests
time series forecasting
variable selection
url https://www.mdpi.com/1999-4893/10/4/114
work_keys_str_mv AT hristostyralis variableselectionintimeseriesforecastingusingrandomforests
AT georgiapapacharalampous variableselectionintimeseriesforecastingusingrandomforests
_version_ 1725664087063396352

Similar Items