Machine learning may not be as good as expected : Evidence from unemployment rate forecasting

46 Pages Posted: 17 Dec 2019 Last revised: 30 Jul 2021

See all articles by Tsung-Wu Ho

Tsung-Wu Ho

National Taiwan Normal University, College of Management

Date Written: July 30, 2021

Abstract

This paper proposes a training framework by rolling k-fold cross-validation to compare forecasting performance of several quantitative methods, mainly standard time series and our pre-selected machine learning methods. Using US unemployment rate, we find that: Firstly, individual machine learning constituents may not perform as good as standard time series; secondly, among on constituent basis, SVM (support vector machine) performs the best, the deep learning (RNN-LSTM) unexpectedly performs the worst; thirdly, forecasting averaging evidence shows that the automatic machine learning (autoML, h2o.ai) performs less than our pre-selected machine learning methods, and the averaged standard time series is better than autoML. We conclude that forecasting averaging is a good way to combine diversified forecasts and a suitable combination of methods depends on the data.

Keywords: Forecasting time series, forecasting averaging, machine learning, training by rolling k-fold cross validation

JEL Classification: C53

Suggested Citation

Ho, Tsung-Wu, Machine learning may not be as good as expected : Evidence from unemployment rate forecasting (July 30, 2021). Available at SSRN: https://ssrn.com/abstract=3496138 or http://dx.doi.org/10.2139/ssrn.3496138

Tsung-Wu Ho (Contact Author)

National Taiwan Normal University, College of Management ( email )

No. 31, Shi-Da Road
Taipei, 10610
Taiwan
+886(2)7734-5624 (Phone)

HOME PAGE: http://web.ntnu.edu.tw/~tsungwu/

Do you have a job opening that you would like to promote on SSRN?

Paper statistics

Downloads
386
Abstract Views
1,374
rank
95,985
PlumX Metrics