Time series forecasting with recurrent neural networks

Time series, such as demand trends, stock prices, and sensor data, is an essential data type in our modern world. Over the years, many models such as Exponential Smoothing and ARIMA are developed to make forecasts on time series. Recently, Recurrent Neural Networks (RNN) is gaining traction in the field of time series forecasting. RNN is a type of specialized neural network tailored towards handling sequential data such as natural language and time series. RNN models such as LSTM networks and GRU networks are widely used in literature. Besides, different feature engineering methods such as CEEMDAN are also tools employed in the literature to improve prediction accuracy. In this paper, we will introduce different models and methods of handling time series and will conduct a comparative case study using the S$\&$P500 index to compare the effectiveness of these models.

Name thesis:degree_name

M.S.

Level thesis:degree_level

Thesis

Discipline thesis:degree_discipline

Industrial Engineering

Grantor

University of Illinois at Urbana-Champaign

Year dc:date

2021

Author dc:creator

• Pan, Zhonghao

Contributors dc:contributor

• Kim, Harrison M

• Time Series
• Recurrent Neural Networks
• LSTM
• GRU
• CEEMDAN

Statement dc:rights

• Copyright 2021 Zhonghao Pan

Language dc:language

en

Handle dc:identifier

http://hdl.handle.net/2142/110479

OAI identifier oai:identifier

oai:www.ideals.illinois.edu:2142/110479