LASSO and attention-TCN: a concurrent method for indoor particulate matter prediction

Area of Science:

• Environmental Science
• Data Science
• Public Health

Background:

• Long-term exposure to indoor air pollution poses significant risks to cardiovascular and respiratory health.
• Existing research predominantly focuses on outdoor air quality, with limited attention to indoor environments.
• Current neural network models for indoor air quality prediction suffer from issues like information loss, high memory usage, and slow processing times due to serial feature input and lack of input variable optimization.

Purpose of the Study:

• To develop a novel, concurrent indoor particulate matter (PM) prediction model.
• To optimize input variables and reduce information loss during model training for improved prediction accuracy and efficiency.
• To identify key environmental factors influencing indoor PM concentrations.

Main Methods:

• A fusion model, LATCN (Least Absolute Shrinkage and Selection Operator - Attention Temporal Convolutional Network), was developed.
• LASSO regression was employed for feature selection from PM datasets (PM1, PM2.5, PM10, PM(>10)) and environmental factors.
• An Attention Mechanism (AM) was used to extract key features by reducing redundant temporal information, followed by a Temporal Convolutional Network (TCN) for parallel forecasting with residual connections to minimize information loss.

Main Results:

• Key environmental factors influencing indoor PM concentration were identified as indoor heat index, indoor wind chill, wet bulb temperature, and relative humidity.
• LATCN demonstrated significant improvements over Long Short-Term Memory (LSTM) and Gated Recurrent Unit (GRU) models.
• LATCN achieved prediction error rate reductions of 19.7%–28.1% (NAE) and 16.4%–21.5% (RMSE), and improved model running speed by 30.4%–81.2%.

Conclusions:

• The LATCN model offers a more accurate and efficient approach to indoor PM prediction.
• Findings provide crucial insights for active prevention of indoor air pollution and inform the development of indoor environmental standards.
• This research lays the groundwork for future innovations in air pollution prevention equipment.