A Comprehensive Evaluation of Statistical, Machine Learning and Deep Learning Models for Time Series Prediction

EasyChair Preprint 6716, version 2

Abstract

How to choose the appropriate model to predict the time series is one of the most prominent activities of temporal data analysis. Empirical evidence is often adopted to select the most suitable model since there is no unified standard for matching data and models. Data characteristics affect model performance to a certain extent and maybe where the factors that determine the balance between prediction accuracy and model complexity are. In this article, Multi-Criteria Performance Measure method considering Mean of Absolute Value of the Residual Autocorrelation was adopted to address this problem. Case studies summarize the limitations and recommendations from the period, trend, stationarity and seasonality of datasets. The results show that the statistical models perform best for datasets with low stochasticity and high trend and seasonality, deep learning models specialize in forecasting fluctuant and long-term time series data, machine learning models could be candidates for datasets that possess numerical characters between the previous two categories. Conclusions could provide suggestions in selecting appropriate models and guide the research community in focusing the effort on more feasible or promising directions.

Keyphrases: Machine Learning Model, Multi-Criteria Performance Measure, deep learning model, statistical model, time-series data prediction

@booklet{EasyChair:6716,
  author    = {Ang Xuan and Mengmeng Yin and Yupei Li and Xiyu Chen and Zhenliang Ma},
  title     = {A Comprehensive Evaluation of Statistical, Machine Learning and Deep Learning Models for Time Series Prediction},
  howpublished = {EasyChair Preprint 6716},
  year      = {EasyChair, 2021}}