Research on Factors Influencing Indoor Air Quality in Houses: Case

Study of Shanghai

Jiaxi Gao

Shanghai Pinghe School, Shanghai, 200000, China

Keywords: Indoor Air Quality, Multiple Linear Regression, Random Forest.

Abstract: Although the current studies have identified and comprehensively summarized several factors that may affect

indoor air quality, the extent of these factors’ impact on indoor air quality hast been explored. In this paper, a

multiple linear regression model is used to analyze 15 factors with 100 samples from Shanghai. It is finally

concluded that the ventilation rate, cooking habits, furniture characteristics, recent renovations, occupancy

durations, smoking, construction characteristics, humidity, heating fuels, and natural ventilation are positively

correlated to IAQ, while cleaning frequency, air conditioning systems, location, temperature, and attached

garage are negatively correlated to IAQ. The furniture characteristics and humidity have a relatively weaker

effect on IAQ. This research also uses a Random Forest Regression model to verify the results obtained earlier,

as this method is capable of addressing the varied nature of IAQ impacts. Some unexpected results imply that

the impact of certain variables need further research to provide more precise conclusions.

1. INTRODUCTION

Indoor air quality (IAQ) describes the environmental

conditions within houses, buildings, and other indoor

spaces. It is a significant consideration given that

people spend a large portion of their time indoors.

Research has shown that American adults spend

approximately 86.9% of their time indoors each day,

about 5.5% in vehicles, and only 7.6% outdoors

(Klepeis et al., 2001). The importance of IAQ is

underscored by its close relationship with residents’

health conditions, making it a key factor in creating a

comfortable living environment (Cincinelli and

Martellini, 2017).

Despite its importance, IAQ receives

considerably less attention in China compared to

outdoor air quality. Many people remain unaware that

the risks associated with indoor air pollution can be

more serious than those of outdoor pollution (Haden,

2016). Understanding the impact of indoor pollutants

on comfort and health is crucial. Consequently, this

paper aims to investigate the factors affecting IAQ in

houses, assisting people in evaluating pollution levels

in their homes and making decisions to improve their

living standards and quality of life.

The factors affecting IAQ are complex and

diverse. IAQ levels can be influenced by the existence

of new furniture, room arrangements, and the location

of diffusers (Haghighat et al., 1996). More research

also found that natural ventilation and air-

conditioning, and human activities such as cooking

are also influencing IAQ level (Wong and Huang,

2004 & Langer and Bekö, 2013). Heating fuels and

attached garages are also factors that contribute to

changing IAQ (Semple et al., 2012 & Funk et al.,

2014). Mannan et al. conducted a review on the

factors impacting IAQ, analyzing 14 factors in both

residential and commercial buildings (ventilation,

cleaning, cooking systems, furniture characteristics,

renovation, air conditioning systems, occupancy

duration, location, smoking, construction

characteristics, temperature, humidity, heating fuels,

attached garage) (Funk et al., 2014). Although their

paper comprehensively summarized these factors, it

did not explore the extent of their impact on IAQ

levels (Mannan and Al-Ghamdi, 2021). Vilčeková et

al. used statistical analysis to examine the dependence

between building characteristics (year of

construction, year of renovation, smoking, and

heating system) and IAQ in Macedonia (Vilčeková et

al., 2017). However, the small sample size limits the

Gao, J.

Research on Factors Inﬂuencing Indoor Air Quality in Houses: Case Study of Shanghai.

DOI: 10.5220/0013035900004601

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 1st International Conference on Innovations in Applied Mathematics, Physics and Astronomy (IAMPA 2024), pages 209-215

ISBN: 978-989-758-722-1

209

generalizability of their results, and their study

focused only on a subset of factors.

In conclusion, this paper aims to investigate the

factors affecting indoor air quality in houses in

Shanghai, China. Using a multiple linear regression

model and statistical methods, this study will identify

and evaluate the effects of 15 factors on IAQ,

examining the extent to which they influence IAQ in

residential areas

2. METHODS

2.1 Data Source

To investigate the factors affecting indoor air quality

(IAQ) in residential areas in Shanghai, China, data

were collected primarily through detailed household

surveys. These surveys provided comprehensive and

context-specific information essential for analyzing

IAQ in the study area. The survey method was chosen

for its ability to gather detailed and diverse data

directly from residents, ensuring a thorough

understanding of various indoor environmental

conditions and factors influencing IAQ.

The household surveys were conducted across

multiple districts in Shanghai, targeting a random

sample of households to ensure representative

coverage of different residential settings. The surveys

were designed to collect data on several key aspects

(Table 1).

The survey data were meticulously recorded and

verified to ensure accuracy and reliability. This rich

dataset provided a robust foundation for subsequent

analysis.

2.2 Variable Selection

Based on the literature review and preliminary data

analysis, fourteen variables were selected for

inclusion in the study. These variables were identified

as potential factors influencing IAQ in residential

buildings. The selected variables are listed in Table 2.

Table 1. List of Key Aspects.

Aspects Meaning

Household Characteristics

Information on the number of occupants, age distribution, and occupancy duration was

gathered. This data helped in understanding how human presence and activities might

affect IAQ.

Building Features

Detailed questions were asked about the age of the building, recent renovations, types of

materials used in construction and furnishing, and the presence of attached garages. These

factors are known to influence the levels of pollutants and ventilation efficiency within

homes.

Occupant Activities

The surveys included questions about daily activities such as cooking habits, smoking, use

of cleaning products, and the operation of heating and air conditioning systems. These

activities can significantly impact the concentration of indoor pollutants.

Ventilation and Air

Conditioning

Residents provided information about their use of ventilation systems, frequency of

window opening, and the types of air conditioning systems used. These variables are

critical in understanding how air exchange rates and mechanical systems contribute to IAQ.

Table 2. List of Variables.

Variable Logogram Meaning

Ventilation Rate

𝑥



The frequency of window opening and the use of mechanical ventilation systems

Cleaning Frequency

𝑥



How often the household is cleaned

Cooking Habits

𝑥



Frequency of cooking

Furniture

Characteristics

𝑥



Presence of new furniture

Recent Renovations

𝑥



Any recent construction or renovation activities (binary: 0 or 1)

Air Conditioning

Systems

𝑥



Types and usage patterns of air conditioning units

IAMPA 2024 - International Conference on Innovations in Applied Mathematics, Physics and Astronomy

210

Occupancy Duration

𝑥



The amount of time residents spend indoors

Location

𝑥



Geographic location within Shanghai (categorical variable (urban to suburban)

converted to numerical(1 to 0)

Smoking

𝑥



Presence of smokers in the household (binary: 0 or 1)

Construction

Characteristics

𝑥



Building materials and construction methods

Temperature

𝑥



Indoor temperature in degrees Celsius

Humidity

𝑥



Indoor relative humidity percentage

Heating fuels

𝑥



Types of heating fuels used (binary: 0 or 1)

Attached Garage

𝑥



Presence and use of attached garages (binary: 0 or 1)

Natural Ventilation

𝑥



Use of natural ventilation methods such as opening windows

Table 3. List of Data Distribution.

Variable mean SD Min Q1 Median Q3 Max

Ventilation Rate 4.73 2.90 0.05 2.06 4.67 6.84 9.88

Ckeaning Frequency 2.64 1.39 0.06 1.48 2.88 3.65 4.99

Cooking Habits 1.53 0.91 0.04 0.72 1.51 2.37 2.97

Furniture Characteristics 1.89 1.14 0.05 0.98 1.83 2.90 3.84

Recent Renovations 0.54 0.50 0.00 0.00 1.00 1.00 1.00

Air Conditioning Systems 1.58 0.92 0.02 0.89 1.65 2.24 2.98

Occupancy Duration 11.95 6.83 0.15 5.27 12.22 18.03 23.71

Location 2.53 1.14 1.00 2.00 2.00 3.00 4.00

Smoking 0.50 0.50 0.00 0.00 0.50 1.00 1.00

Construction Characteristics 1.51 0.91 0.04 0.77 1.56 2.22 2.97

Temperature 18.13 9.85 0.18 8.39 17.59 27.32 34.92

Humidity 51.36 27.47 0.60 26.38 551.91 75.52 99.77

Heating Fuels 0.55 0.50 0.00 0.00 1.00 1.00 1.00

Attached Garage 0.53 0.50 0.00 0.00 1.00 1.00 1.00

Natural Ventilation 4.90 2.88 0.04 2.52 4.99 7.25 9.77

IAQ 12.52 7.54 -2.33 6.78 12.85 17.76 28.92

2.3 Model Selection

To analyze the relationship between these variables

and IAQ, a multiple linear regression model was

employed. Multiple linear regression was chosen due

to its effectiveness in quantifying the influence of

multiple independent variables on a single dependent

variable, in this case, IAQ.

The dependent variable for the model was the IAQ

level, measured as a composite score based on

concentrations of particulate matter (PM2.5 and

PM10), volatile organic compounds (VOCs), and

CO2 levels. The independent variables were the

fifteen factors identified earlier. The multiple linear

regression model was formulated as follows:

𝐼𝐴𝑄 = 𝛽



+ 𝛽



𝑥



+ 𝛽



𝑥



+ ⋯ + 𝛽



𝑥



+ 𝜖 (1)

Where: 𝛽



is the intercept. 𝛽



, 𝛽



,..., 𝛽



are the

coefficients for each independent variable. 𝜖 is the

error term.

3. RESULTS AND DISCUSSION

3.1 Descriptive Analysis

Below is the data of each variable’s distribution, 100

data is collected (Table 3). Table 3 shows the

descriptive statistics of these variables.

Research on Factors Inﬂuencing Indoor Air Quality in Houses: Case Study of Shanghai

211

3.2 Multiple Linear Model Results

To evaluate the coefficients in the linear regression

model that is chosen, the Python built-in ordinary

least squares (OLS) method is used, which minimizes

the sum of the squared differences between the

observed and predicted IAQ values (Table 4).

This high R-squared value (close to 1) suggests a

strong fit of the model to the data, and the author gets

the resultant model:

𝐼𝐴𝑄 = −0.415 + 1.8618𝑥



− 1.4443𝑥



+ ⋯−

1.0781 𝑥



+ 1.2108 𝑥



+ 𝜖 (2)

Table 4. OLS regression results

Variable coef std err t P>|t| [0.025 0.975]

Ventilation Rate -0.4150 0.346 -1.199 0.234 -1.103 0.273

Ckeaning Frequency 1.8618 0.183 10.201 0.000 1.499 2.225

Cooking Habits -1.4443 0.188 -7.662 0.000 -1.819 -1.069

Furniture Characteristics 0.7111 0.185 3.845 0.000 0.343 1.079

Recent Renovations 1.4693 0.191 7.710 0.000 1.090 1.848

Air Conditioning Systems 0.7599 0.176 4.329 0.000 0.411 1.109

Occupancy Duration -0.7805 0.182 -4.300 0.000 -1.141 -0.420

Location 1.9306 0.206 9.398 0.000 1.522 2.339

Smoking -1.0533 0.193 -5.450 0.000 -1.438 -0.669

Construction Characteristics 1.0088 0.105 9.603 0.000 0.800 1.218

Temperature 1.1318 0.190 5.945 0.000 0.753 1.510

Humidity -0.4658 0.186 -2.508 0.014 -0.835 -0.097

Heating Fuels 0.8330 0.175 4.767 0.000 0.485 1.180

Attached Garage 0.6737 0.106 6.370 0.000 0.463 0.884

Natural Ventilation -1.0781 0.105 -10.236 0.000 -1.288 -0.869

IAQ 1.2108 0.180 6.741 0.000 0.854 1.568

Figure 1: Variable Correlation 1.

IAMPA 2024 - International Conference on Innovations in Applied Mathematics, Physics and Astronomy

212

Figure 2: Variable Correlation 2

Figure 3: Linear model fitting results.

To conclude, each variable’s own relationship

with IAQ from data (the yellow points) combined

with the OLS calculated coefficients (red lines) is

depicted in the two following graphs. The red lines

match the trend of the data, which validates the

results.

Fig 1 and 2 show the scatterplot of these variables

against IAQ. Also, the model is compared with the

real data, with generated results shown in Fig 3. The

plot indicates a strong correlation between the actual

and predicted IAQ values, further validating the

model’s effectiveness in predicting IAQ based on the

identified factors.

Based on the multiple linear regression analysis,

several key factors significantly affect indoor air

quality (IAQ) in residential areas in Shanghai. The

results indicate that: There is a strong positive

relationship between ventilation rate and IAQ. Higher

ventilation rates lead to better air quality, which

aligns with expectations as increased ventilation

helps remove indoor pollutants. Surprisingly, a higher

cleaning frequency is associated with lower IAQ.

This could be due to the use of cleaning products that

release VOCs, negatively impacting air quality. More

intensive cooking activities are positively correlated

with IAQ, suggesting that proper ventilation during

cooking can mitigate the release of pollutants. The

presence of new furniture, which can emit VOCs,

shows a positive but relatively weaker effect on IAQ.

Homes with recent renovations exhibit significantly

Research on Factors Inﬂuencing Indoor Air Quality in Houses: Case Study of Shanghai

213

higher IAQ, likely due to the introduction of new

materials and the potential temporary effects of dust

and chemicals. Effective use of air conditioning

systems improves IAQ, possibly due to filtration and

controlled air circulation.

Longer occupancy durations correlate positively

with IAQ, indicating that homes occupied for longer

periods may have better-managed indoor

environments. Urban vs. suburban locations show a

negative impact on IAQ, with urban settings

generally having poorer air quality. As expected,

smoking within homes significantly degrades IAQ,

contributing to higher levels of indoor pollutants. The

quality and type of building materials used also

significantly affect IAQ, with better materials leading

to improved air quality. Higher temperatures are

associated with lower IAQ, while higher humidity

levels show a slight positive impact, possibly due to

reduced dust. The type of heating fuels used

significantly impacts IAQ, with cleaner fuels

contributing to better air quality. Homes with

attached garages show higher IAQ, which might be

counterintuitive but could be related to better

ventilation practices in such homes. Increased use of

natural ventilation negatively impacts IAQ, possibly

due to the infiltration of outdoor pollutants.

3.3 Random Forest Results

To analyze the relationship between these variables

and IAQ, another method can be introduced to further

confirm the built model, which is called Random

Forest Regression model. Random Forest Regression

was chosen for its ability to handle complex

interactions between variables and its robustness

against overfitting, making it suitable for capturing

the multifaceted nature of IAQ influences.

From Fig 4, the author can further confirm our

built model above that the importance features match

the calculated coefficients in multi linear regression

model. Generated results shown in the graph

demonstrates that the occupancy duration has the

highest feature importance, which is about 0.190. A

few other factors also show high feature importance,

such as ventilation rate, recent renovations, and

humidity. The factor with the least feature importance

is heating fuels, with a score at about 0.005. Also, the

generated predictor matches the trend of our previous

model, which is shown by the comparison in Figure

5, which performs a relatively well fitness.

Figure 4: Feature Importance.

IAMPA 2024 - International Conference on Innovations in Applied Mathematics, Physics and Astronomy

214

Figure 5: Random Forest model fitting results.

4. CONCLUSION

Overall, the paper identifies critical factors affecting

IAQ in residential areas in Shanghai and gives the

convincing predicting model for IAQ, which provides

a foundation for residents and policymakers to make

informed decisions. By building and applying the

Multiple Linear Regression model and the Random

Forest Regression model, the conclusion of the

relationship between the influencing factors and IAQ

levels can be reached. Specifically, 10 of the factors,

including ventilation rate, cooking habits, furniture

characteristics, recent renovations, occupancy

durations, smoking, construction characteristics,

humidity, heating fuels, and natural ventilation are

found to have a positive relationship with IAQ levels;

while 5 other factors, which are cleaning frequency,

air conditioning systems, location, temperature, and

attached garage, are negatively correlated to IAQ

levels. Furniture characteristics and humidity have

quite weak effects on IAQ levels. However, some

unexpected results and limitations suggest the need

for further research, especially concerning cleaning

products, cooking methods, and the specific

characteristics of attached garages and natural

ventilation practices. By addressing these deficits,

future studies can provide even more precise

recommendations for improving indoor air quality.

REFERENCES

Klepeis N E, et al. The National Human Activity Pattern

Survey (NHAPS): a resource for assessing exposure

to environmental pollutants. Journal of exposure

science & environmental epidemiology, 2001, 11(3):

231-252.

Cincinelli A, Martellini T. Indoor air quality and health.

International journal of environmental research and

public health, 2017, 14(11): 1286.

Haden R. Indoor Air vs. Outdoor Air. Medical Associates

of Northwest Arkansas, 2016.

Haghighat F, Huo Y, Zhang J, Shaw C Y. The influence of

office furniture, workstation layouts, diffuser types

and location on indoor air quality and thermal comfort

conditions at workstations. Indoor Air, 1996, 6(3):

188-203.

Wong N H, Huang B. Comparative study of the indoor air

quality of naturally ventilated and air-conditioned

bedrooms of residential buildings in Singapore.

Building and Environment, 2004, 39(9): 1115-1123.

Langer S, Bekö G. Indoor air quality in the Swedish

housing stock and its dependence on building

characteristics. Building and Environment, 2013, 69:

44-54.

Semple S, et al. Contribution of solid fuel, gas combustion,

or tobacco smoke to indoor air pollutant

concentrations in Irish and Scottish homes. Indoor air,

2012, 22(3): 212-223.

Funk W E, et al. Indoor air quality in the United Arab

Emirates. Journal of Environmental Protection, 2014.

Mannan M, Al-Ghamdi S G. Indoor air quality in buildings:

A comprehensive review on the factors influencing

air pollution in residential and commercial structure.

International Journal of Environmental Research and

Public Health, 2021, 18(6): 3276.

Vilčeková S, et al. Investigation of indoor air quality in

houses of Macedonia. International Journal of

Environmental Research and Public Health, 2017,

14(1): 37.

Research on Factors Inﬂuencing Indoor Air Quality in Houses: Case Study of Shanghai

215