PM2.5 Forecasting Model Using a Combination of Deep Learning and Statistical Feature Selection

Endah Kristiani,Kieu Lan Phuong Nguyen,Ting-Yu Kuo,Kai-Chih Pai,Chao-Tung Yang,Chin-Yin Huang

doi:10.1109/access.2021.3077574

Abstract

This paper proposed a PM 2.5 forecasting model using Long Short-Term Model (LSTM) sequence to sequence combined with the statistical method. Correlation Analysis, XGBoost, and Chemical processed are used as the methods to select the essential features. The air pollution data is extracted from Taiwan Environmental Protection Agency (EPA) for the Taichung City dataset in 2014–2018. The study points out that chemical processed model of particulate matter 10 micrometers or less in diameter (PM10), Sulfur Dioxide (SO2), and Nitrogen Dioxide (NO2) have the highest accuracy or lowest Root Mean Square Error (RMSE) and more short training and testing time among the other models. The chemical processed model of PM10, SO2, and NO2 (model B) has the highest accuracy (lowest RMSE), approximately 1 point lower RMSE values, and the shortest training and testing period among the other models. Furthermore, RMSE calculations based on the stations reveal that training with the entire station dataset has a 3 point higher RMSE value than training with each station dataset.

Highlights

I NTERNET of Things (IoT) is an interconnection of various instruments, networks, techniques, and human resources for a common purpose
NO2, SO2; (3) model C: using Long Short-Term Model (LSTM) seq2seq for PM10, O3, AMB TEMP; (4) model D: using LSTM seq2seq for top 5 parameters selected by correlation analysis; and (5) model E: using LSTM seq2seq for top 5 parameters selected by feature selection
This paper demonstrated the PM2.5 forecasting model using Long Short-Term Model (LSTM) seq2seq combined with the statistical method

Summary

Introduction

I NTERNET of Things (IoT) is an interconnection of various instruments, networks, techniques, and human resources for a common purpose. Data analytics is used to examine large and small information sets with different data characteristics to draw significant findings and practical insights. These findings are generally in the form of trends, patterns, and statistics that support businesses in the proactive use of information for efficient decisionmaking. A more complicated form of sequence prediction problem takes a sequence as input and requires a sequence prediction as output [30]. These are referred to as sequence-to-sequence prediction (seq2seq) problems. NOx, SOx, and O3 and temperature are critical factors affecting PM2.5 concentration in the atmosphere

Methods

Results

Conclusion