=Paper=
{{Paper
|id=Vol-3940/AISD-2024_Paper_3
|storemode=property
|title=Insights from Data Science and Machine Learning: Understanding Global Air Quality Dynamics
|pdfUrl=https://ceur-ws.org/Vol-3940/AISD-2024_Paper_3.pdf
|volume=Vol-3940
|authors=Sai Shradha V,Kukatlapalli Pradeep Kumar,Vijaya Padmanabha
|dblpUrl=https://dblp.org/rec/conf/aisd/VKP24
}}
==Insights from Data Science and Machine Learning: Understanding Global Air Quality Dynamics ==
https://ceur-ws.org/Vol-3940/AISD-2024_Paper_3.pdf
Insights from Data Science and Machine Learning:
Understanding Global Air Quality Dynamics
Sai Shradha V 1, , Kukatlapalli Pradeep Kumar2, and P Vijaya 3,
1,2 Department of Computer Science and Engineering, Christ University, Bangalore, India.
3 Department of Mathematics and Computer Science, Modern College of Business and Science, Muscat, PC 133, Sultanate of
Oman.
Abstract
Air pollution is one of the growing global phenomena which is affecting all living organ-isms on the planet
and is one of the major climatic crisis that has to be addressed efficiently in a sustainable manner. Air
pollution is mainly rising due to the increasing amount of ozone, carbon monoxide and Particulate Matter
2.5 because of various reasons including emission of these gases from industries, burning of fossil fuels,
increasing number of aero-sols in the air which has ultimately led to global warming and ozone layer
depletion. This research paper mainly aims to focuses on analysis and interpretation of the impact of in-
creasing levels of Ozone, Carbon monoxide, Particulate Matter 2.5 and other pollutants which contributes
to global air pollution trends across various major cities around the world including Indian cities. Various
hypothesis tests had been performed such as One-way Analysis of Variance tests, to demonstrate the
relationship between the impact of contaminants, primarily carbon monoxide, ozone, Particulate Matter 2.5
on the overall Air Quality index various major cities around the world. The effect of the major pollutant on
the overall Air Quality Index of the most of the countries has been identified and various machine learning
algorithms like K-Nearest Neighbor, Logistic regression, LSTM, Random forests have been implemented to
compare and analyze the impact of the air pollutants. This re-search also contributes to United Nations
Sustainable development goals on Climate action to study and reduce the increasing global warming which
leads to increase in temperature, melting of ice caps day by day.
Keywords
Data Analytics, Global Warming, Air Quality Index, Carbon monoxide, Statistical tests 1
1. Introduction
Human wellbeing is greatly impacted by the indoor conditions, since ninety per-cent of people spend
more than ninety percent of their time indoors. Every year, indoor air pollution, also known as indoor
air pollution, causes 3.8 million deaths. It can be caused by the activities of occupants, including
cooking, smoking, using electronics, and emitting materials from buildings. Due to shifting life-styles
and urbanization, research on air quality control has moved from outdoor to indoor settings. Human
health may be adversely affected by decreased IAQ, which may result in illnesses linked to buildings.
Approaches for regulating and improving IAQ, as well as monitoring systems, are crucial. This paper
offers a thorough analysis of the main sources of INDOOR air pollution, control tactics, health
implications, health problems, and trends in surveillance and control Finding the sources of air
pollutants is essential to managing indoor air quality effectively.
The overall condition of the outside air, the activities of people inside buildings, and the
furnishings, appliances, and building materials are among the primary variables that affect indoor air
quality Given that pollutants can travel from the outside into the interior, it is well known that the
degree of air sealing in constructions and the quantity of pollutants outdoors have a significant impact
on the condition of indoor air (indoor air quality). The contaminants outside include Particulate
matter, also called particulates (Particulate Matter), dioxide of carbon (CO2), Sulphur oxides (SO2),
AISD-2024: Second International Workshop on Artificial Intelligence: Empowering Sustainable Development, October 2, 2024,
co-located with the Second International Conference on Artificial Intelligence: Towards Sustainable Intelligence (AI4S-2024),
Virtual Event, Lucknow, India.
sai.shradha@btech.christuniversity.in (Sai Shradha V); kukatlapalli.kumar@christuniversity.in (Kukatlapalli Pradeep
Kumar); pvvijaya@gmail.com (P Vijaya)
0009-0000-0661-8088 (Sai Shradha V); 0000-0002-8893-4312 (Kukatlapalli Pradeep Kumar); 0000-0002-2117-9543 (P
Vijaya)
2024 Copyright for this paper by its authors. Use permitted under Creative Commons License Attribution 4.0 International (CC BY 4.0).
CEUR
ceur-ws.org
Workshop ISSN 1613-0073
Proceedings
oxides of carbon (CO), oxides of nitrogen (NO2), and the products of combustion are among the
emissions that are present in in-door air setting. Most importantly, ventilation system design and
operation have a significant effect over the indoor air quality. Air circulation helps keep up good
indoor air quality and foster healthy conditions by exchanging stagnant indoor air with fresh outdoor
air. There are several advantages to running a building's ventilation system [1], such as: (a) supplying
clean air and oxygen that people require to breathe; (ii) bringing indoor air pollution levels down to
levels that fall within short-term exposure limits for dangerous contaminants; and (iii) eliminating
vapors and Odors. Exhaustive Literature review is provided with concepts to biological pollutants,
Sulphur and Nitrogen elements. This is followed by Results and Discussions which includes Data
Visualization, Statistical tests and machine learning observations.
1.1. Objectives
The main objective of this research is to identify the various air pollutants in the atmosphere and
how to contributes to the decrease in Air Quality Index all over the world. Various tests and analysis
have mainly been done as to observe the relationship as well as the influence of these air pollutants
on the Air Quality Index of the atmosphere. Hence various machine learning algorithms have been
chosen specially to identify the impact of Ozone on the atmosphere, which indirectly contributes to
overall Air Quality Index value of a place. After various statistical analysis it has been found out the
significance of ozone on the Air Quality , hence various algorithms have been extensively performed
to under-stand its influence. The accuracy level of the algorithms is also clearly mentioned , if which
the highest is the random forest with which we can directly identify he relation and the dependence
of the major pollutant Ozone on atmospheric as well as climatic conditions.
2. Literature Review
Yang and Zhao's review article from 2023 offers a thorough examination of the air quality models
uses. After reviewing 212 pa-pers released since 2010, the authors found that, whereas CMAQ, WRF-
Chem, and NAQPMS exhibit less definite biases, models like GEOS-Chem and CAMx frequently
overestimate O3 concentrations. The main causes of long-
anthropogenic emissions, especially those from industry and traffic. Short-term pollution episodes
are increasingly influenced by meteorological factors. To increase simulation accuracy, the scientists
stress the need for better emission inventories and model methods. Additionally, they talk about the
-
in areas with low VOC concentrations. The subsequent review primarily addresses the different types
of air pollutants, such as particulate matter and aerosols like nitrogen oxides, Sulphur dioxide, carbon
oxides, and ozone which contributes to the global air pollution.
2.1. Particulate matter
A carbon-based particle connected to reactive metals and organic com-pounds is called a
particulate matter . Due to its potential for inhalation, which can harm the heart and lungs, it poses
a serious health risk. Indoor activities and outdoor particles are two reasons why indoor levels of
frequently surpass outdoor ones. A few of the factors that cause indoor are cooking, burning fossil
fuels, smoking, using machines, and engaging in recreational activities at home. Because Particulate
Matter, a particulate matter derived from the burning of fossil fuels, can enter small airways and
alveoli, it presents a greater risk to health. The primary sources of interior particulate matter come
from cooking and cigars smoking; cleaning activities contribute significantly less.
One quarter of indoor concentrations are caused by other human activities. In every city but
Bengaluru, Particulate Matter 2.5 and Aerosol optical depth show a decreasing trend from January to
May. Although there was more daily variability and less steep trends, satellite-based and Aerosol
optical depth revealed trends, satellite-based and Aerosol optical depth revealed trends that were
comparable to surface measurements. Bengaluru saw a significant de-crease in Particulate Matter 2.5
concentrations during the lockdown, with daily readings as low as thirteen and a half microgram
high levels of particulate matter in the month of January, but in 2020, concentrations of Particulate
Matter 2.5 were roughly fourteen percent lower than they were prior to the lock down [1].
2.2. Nitrogen Oxides
Two of the main nitrogen oxides linked to sources of combustion like heaters and cooking stoves
are nitrogen oxides (NO) and nitrogen dioxide (NO2). The local sources and sinks of nitrogen have
an impact on the ambient concentrations of NO and NO2. Under ambient conditions, NO2 is regarded
as a primary pollutant; however, NO2 quickly oxidizes to form NO2. Nitrous acid (HNO2 ) is a
powerful oxidant that is produced when NO2 com-bines with water. It is a common indoor pollution.
Both indoor and outdoor sources can raise or lower indoor NO2 levels, and the separation between
buildings and roads has a major impact. [2]
Cities like Bangalore, Delhi, the city of Kolkata, and Pune, Maharashtra have high seasonality in
their nitrogen dioxide (NO2) concentrations, which de-cline from January to May. The main causes
of this seasonality are variations in mixing layer depth and lifetime. All six cities saw a notable
decrease in NO2 concentrations after the imposition of stringent security precautions in March 2020,
with Delhi exhibiting the greatest decrease. These declines are similar to those observed in Chinese
and European cities and correlate with the falls in tropospheric nitrogen dioxide segments detected
by Ozone monitoring instrument [3].
2.3. Sulphur Dioxide
The most prevalent and important gas among the Sulphur oxides in the atmosphere is Sulphur
dioxide (SO2), which is mostly produced by utilizing fossil fuels. It mixes with Particulate Matters and
aerosols to create a complex mixture of unique air. Another major source of indoor SO2 is outside air.
Oil furnaces, aired gas-powered devices, smoke from cigarettes, kerosene heaters, coal or wood
stoves, and other indoor sources are among those that release SO2. Because SO2 is reactive, indoor
surfaces can easily absorb it, and indoor levels of the gas are typically lower than outdoor levels.
Buildings usually have hourly SO2 concentrations of less than 20 parts per billion, and human
exposure to SO2 predominantly impacts the way we breathe by means of inhalation. Industries, which
includes thermal electric power plants, is responsible for more than eighty per-cent of SO2 emissions;
through the lockdown, average Sulphur dioxide amounts in Chennai and Delhi dropped somewhat.
Pune saw a fairly stable two-year peri-od whereas Bengaluru and the city of Kolkata experienced
moderate increases of thirty-one percent and forty-two percent, respectively. During the lockdown
period of time, Mumbai had a significant spike in mean SO2 concentrations (81percentage) with levels
2.4. Carbon Monoxide gas
Carbon monoxide (also known as CO )is an odorless, colorless gas that is mainly out over processes
of combustion like heating or cooking. Furthermore, it could enter indoor spaces through outside air.
Gas space heaters, unvented kerosene heaters, leaky burners and furnaces, gas water heaters, wood
cookers, fireplaces, stoves with gas, generators and cigarette smoke are the primary sources of indoor
atmospheric carbon monoxide (CO) emissions. The typical Carbon monoxide. Concentrations vary
from half to five parts per million in a structure without gas stoves. Thus, when the building is close
to a gas stove, the level of carbon dioxide can be as high as 30 parts per million. Carbon monoxide
(CO) being exposed can have detrimental effects on one's health, including effects on the nervous
system and the heart. It can also increase the risk of death [5].
2.5. Ozone
Cities' concentrations of ozone (O3) varied daily and did not exhibit a steady trend over the course
of the analysis. Except for Kolkata and Pune, concentrations were, nevertheless, somewhat lower than
the mean of the preceding year (2017 19). While Pune possessed higher concentrations up until the
third week of March 2020, throughout 2020, Kolkata recorded ozone concentrations that were higher
than the baseline. It is reasonable to anticipate a minor rise in volatile organic compound (VOC)
emissions, particularly in circumstances akin to Covid-19 where stringent community disinfection
protocols are implemented to curb the disease's transmission. Nonetheless, NOx emissions in Kolkata
decreased, which is in line with a forty seven percent de-crease in NO2, but it's possible that a
chemical regime limited by VOCs enhanced the formation of O3. When exposed to three thousand
parts per mil-lion of carbon dioxide (CO2), headaches become more intense, drowsiness, fatigue, and
concentration problems occur.
2.6. Biological pollutants
Biological allergens including pollen, cockroaches, house dust, and animal dan-der are examples
of biological pollutants that can be found in indoor environments. Moreover, bacteria, fungus, and
viruses are examples of microorganisms that contribute to biological pollutants indoors. These
allergens come from fungi, animals, or insects and can trigger allergic reactions when certain
are a few examples of indoor allergen sources in addition to external ones. Sensitivity, infections of
the respiratory tract, breathing allergic diseases, and wheezing fits can all result from viral and
bacterial exposure. Exposure to bacteria and viruses indoors can lead to a variety of adverse health
is-sues , which includes infectious and not transmissible ones. Following the guide-lines for managing
asthma can help to lessen the effects, but taking global action is necessary to reduce exposure and
improve asthma outcomes. It is necessary to conduct more research to look into dual or various
exposures and to find respiratory disease patterns that could increase an individual's vulnerability to
atmospheric pollution [7-9].
3. Problem Statement
There are innumerous number of pollutants present in the atmosphere which has severe effect on the
atmosphere. Hence it is very necessary to identify and analyze the pollutants present in the
atmosphere. The major problem is to identify the pollutants that have the most impact on the Air
Quality Index of different regions.
A wide range of pollutants, such as Particulate Matter , carbon dioxide , sulfur oxides , carbon
monoxide , nitrogen oxides , and ground-level ozone , are contributing to a becoming complex and
interconnected web of environmental degradation. Particulate Matter, using its fine particles, not only
impairs air quality but additionally leads to soil and water contamination, influencing biodiversity
and ecosystem functions. Air pollution has emerged as one of the most pressing global environmental
challenges with profound implications for ecosystems, climate, and human health. Rising carbon
dioxide levels are hastening climate change, causing extreme weather, disturbances to natural
habitats, and changes in temperature patterns. Acid rain's primary ingredients, sulfur and nitrogen
oxides, are destroying built environments, aquatic systems, and soils. This has a domino impact on
cultural heritage and food security. Because ground-level ozone is created by photochemical
interactions between NOx and volatile organic com-pounds (VOCs), it can harm flora, lower crop
yields, and have an adverse effect on the health of forests. Meanwhile, by encouraging the continued
presence of other greenhouse gases, carbon monoxide's impact on the atmospheric composition
exacerbates climate change. When combined, these pollutants are causing a number of environmental
problems, such as ocean acidification, desertification, loss of biodiversity, and diminished natural
resilience. Fighting global air pollution necessitates a coordinated and diversified response because
of the interconnection of these pollutants and their diverse sources, which range from farming and
urban development to industrial emissions and transportation. The overall environmental effects of
these pollutants will be examined in this study, with particular attention paid to their
interconnections, and accumulated effects on climate and ecosystems, mainly Air Quality Index. In
doing so, it hopes to further knowledge of the worldwide air pollution situation and provide guidance
for policies aimed at reducing its far-reaching effects.
4. Methodology
Here we conceptualize various graphs including bar graphs, histogram, scatter plots, statistical tests
like One-Way ANOVA tests, and machine learning model implementations. We have collected Global
air pollution dataset from an open source which consists of 23464 rows of various cities air quality
data with variables like Country, City, Air Quality Index Value, Air Quality Index Category, Carbon
Monoxide Air Quality Index Value, Carbon Monoxide Air Quality Index Category, Ozone air quality
index Value, Ozone Air Quality Index Category, NO2 Air Quality Index Value, NO2 Air Quality Index
Category, Particulate Matter 2.5 Air Quality Index Value, Particulate Matter 2.5 air quality index
Category.
4.1. Dataset Quality, Completeness and potential biases
This study's dataset, which focuses on global air pollution, includes detailed geo-located data for
major pollutants like carbon monoxide , nitrogen dioxide , ozone , and particulate matter , along with
corresponding Air Quality Index values and classifications for cities worldwide. The dataset is useful
for examining pollutant distributions and trends in air quality, but in order to make sure that the
conclusions are reliable and correct, it is crucial to talk about its completeness, quality, and any
potential biases.
The fact that the dataset was sourced from an open-source platform has ad-vantages and
disadvantages. Open-source datasets are helpful for extensive analysis because they are generally
widely available and span a large geographic area. However, the manner in which the data were
gathered, handled, and pre-served may have an impact on the data quality. The overall
trustworthiness of the dataset may be impacted by, among other things, disparities in monitoring
equipment, inconsistent data recording techniques, or different data collection intervals between
regions [10].
The dataset may have biases due to a number of causes. For example, the majority of monitoring
stations for air quality are situated in or close to metropolitan areas, which causes an
overrepresentation of areas with greater amounts of pollutants. The results could be skewed overall
if rural areas or locations with fewer monitoring resources are underrepresented. A number of pre-
processing procedures, including data cleaning, were carried out to address problems with data
quality, accuracy, and biases. Outliers and erroneous values were found and either fixed or eliminated
in order to increase the confidence of the results. In order to ensure more dependable results and
reduce noise in the dataset, this step was crucial.
4.2. Dataset Quality, Completeness and potential biases
In order to validate the statistical analysis, thorough explanations of the ANOVA and the cross-
tabulation findings must be given. An ANOVA may reveal significant differences in AQI when the
AQI is categorized by specific pollutant levels, for example, while a cross-tabulation may reveal
correlations between categorical variables like AQI categories and pollutant thresholds. It will support
the discussion of these findings in the context of machine learning outputs, highlighting the
complementary nature of both approaches in highlighting significant factors influencing air quality
changes. By highlighting these statistical results, the paper's conclusions about the effects of various
pollutants, particularly ozone, on AQI are strengthened [11].
The following section gives an insight into and visualization of the impact, frequency and
effect of various pollutants on the environment through Pie charts, bar graphs, histograms, scatter
plots and null hypothesis tests like Pearson correlation, One-way ANOVA test and Chi-squared tests.
Figure 1: Histogram of overall air quality index of the cities
The above Fig. 1. Shows the frequency distribution of histograms of Air Quality Index of various
metropolitan cities of different countries throughout the world acquired from the dataset
Variable taken refers to the different categories of the air quality based on air quality index values.
The variable taken here refers to the air quality index value of Carbon Monoxide and their frequency.
Figure 2: Bar graph of air quality index value of carbon monoxide of the city
The above Fig. 3. Shows bar graph depicting the frequency of Air Quality index value of carbon
monoxide over various cities present as major air pollutant in the atmosphere.
A. Scatter Plot
X-axis Variable is General the air quality index reading for the city.
Y-axis Variable is the city's carbon monoxide Air Quality Index value
Figure 3 quality index value and its
carbon monoxide the air quality index value
The above Fig. 4. Shows a scatter plot on the ratio of Carbon Monoxide to the Overall
Air Quality Index of major cities throughout the world. Histogram of overall air quality
index of the following section depicts various statistical tests including various null
hypothesis tests performed on respective parameters which gives the relation between
various variables and values associated with them [10].Here the skewness and kurtosis
analysis are done on the Overall Air Quality Index value of the city .Since, the value of
kurtosis is, 2.935+0.281=3.216, where it is greater than 3, it is a leptokurtic curve , when
the respective graph is plotted city.
Table 1
Correlation Among the Total Air Quality Index and Air Quality Index Value Of Carbon Monoxide
Pearson Correlation Asymp. Sig. (2-sided)
Overall air
quality index .144
0.984295061448353
value of cities
Air Quality
Index Value of 300 .103
Ozone of The
City
The above Table 1. Shows the Pearson correlation between two variables such as Overall Air Quality
Index value of the city and Air Quality index value of Ozone of the cities which has a positive
correlation of 0.984295061448353The null hypothesis tests are performed on the on the city's overall
air quality index and the carbon monoxide index: Firstly, Chi-Square Tests ,and One-Way ANOVA
tests.
Table 2
Chi-Square Tests on the overall air quality index value and Carbon Monoxide air quality index value
Value df Asymp. Sig. (2-sided)
Pearson Chi-
3.878 2 .144
Square
Likelihood Ratio 4.537 2 .103
The above Table 2. depicts the Chi-Square tests on the two variables, the Air quality index category
of Carbon Monoxide and the Air quality index cate-gory of Nitrogen Dioxide.
The p-value is 0.144 which is greater than 0.05 shows that there is no significant association between
air quality index category of Carbon Monoxide of the city and air quality index category of Nitrogen
Dioxide of the city. One-way Anova Dependent list: Overall Air Quality Index value of the city Fac-
tor: air quality index category of the city (Categorical Variable).
Table 3
One-way ANOVA on the overall air quality index value and Carbon Monoxide air quality index value
Total Squares df Sig. (2-sided)
Between
565.729 1 .011
Groups
Total 749.429 6
The above Table 3. shows the One-way ANOVA Test between Overall Air Quality Index value of the
city and Air Quality Index Category of the city which is a categorical variable having categories good,
bad, hazardous etc. Dependent list taken here is Overall Air Quality Index value of the city and the
Factor is Air Quality Index category of the city.(Categorical Variable).
5. Results
5.1. Machine Learning model perceptive
The capacity of the machine learning models Logistic Regression, Random Forest (RF), LSTM, KNN,
and KNN to identify intricate patterns and correlations in the data on air pollution led to their
selection. Random Forest is an efficient method for determining the most significant contaminants
influencing AQI since it can handle nonlinear interactions well and provides insights into feature
relevance. The ability of LSTM networks to describe temporal connections is particularly helpful
because air quality data is time-series in nature. Because of its ease of use and ability to capture local
trends, K-Nearest Neighbors (KNN) was added [12]. These characteristics can be important when
analyzing the effects of particular contaminants in different locations. Even though it's
straightforward, the benchmark model for binary classification applications, such as classifying AQI
levels, is logistic regression. The amalgamation of these models facilitates an all-encompassing
examination of temporal patterns and pollutant interactions, augmenting the comprehensive
comprehension of ozone's impact on AQI. By performing various correlation analysis, it has been
found that there is a very positive correlation between Ozone air quality index Values , Particulate
matter and the Overall air quality index value of the city which is 0.984295061448353
Various Machine Learning Algorithms are implemented such as K-Nearest Neighbor, Logistic
Regression , Long short-term memory and Random Forests.
Random Forests:
Training variables is ozone air quality index Values and Particulate matter
Testing variables is Overall air quality index value
Mean Absolute Error obtained is 0.01 degrees and Accuracy level 99.98 percent-age
Figure 4: The exponentially increasing graph between the actual versus predicted
values.
The above Figure 4. Shows the exponential growth in the accuracy of the implementation of the
Random Forests algorithms based on the features of Predicted versus Actual values.
Table 4
Result and Accuracy scores of the dataset implemented using KNN
Heading level Values
Baseline K-Nearest Neighbors 0.83 4693
accuracy 0.85 0.46 0.44 4693
macro avg 0.45 0.46 0.44 4693
weighted avg 0.86 0.85 0.85 4693
Mean Square Error for Nearest 13.726
Neighbor
Root Mean Square Error 3.705
Baseline K-Nearest Neighbors 0.785
Scores(Cross validate) for K-Nearest [0.73086144 0.72638645
Neighbors model 0.72234655]
The above Table 4. Shows brief results of the implementation of K-Nearest Neighbor algorithm on
Ozone air quality index Values, Particulate matter and Overall air quality index Value of the cities.
Figure 5: The exponentially increasing graph between the accuracy levels and the cross-validation
score graph is being shown as a result of the K-Nearest Neighbors
The above figure 5 depicts the increasing graph between the accuracy and cross
validation values.
Long short-term memory was implemented and obtained Test Accuracy is: 0.3356062173843384.
Figure 6: The graph shows the model accuracy obtained by implementing Long short-term memory
algorithm.
The above Figure 6 shows the model accuracy rate obtained by implementing Long short-
term memory algorithm,
Logistic Regression implemented with the same testing and training variables and the Test
accuracy is 0.04005966332836139 and Train accuracy is 0.041822056473095365.
Figure 7: The graph shows the impact various pollutants like Ozone, Particulate Matter 2.5 , Nitrogen
Dioxide, Carbon Monoxide on the overall Air Quality Index
The above Figure 7 shows the impact and the effect of major pollutants like Ozone, Particulate Matter
2.5, Nitrogen Dioxide, Carbon Monoxide on the overall Air Quality Index on various cities around
the world thus shows the impact of the pollutants on the health of the people in various ways
Table 5
Result and Accuracy scores of the dataset implemented using KNN
Model Accuracy level
Random Forests. 0.99
K-Nearest
0.83
Neighbor
Logistic
0.040
Regression
Long short-term
0.3356
memory
6. Conclusion
The two types of pollution, such as indoor and outdoor air pollution, are briefly introduced to us in
this paper. Also, the effect of air pollution on people affected by Covid-19 Literature review gives us
insight on various air pollutants ranging from aerosols like Particulate Matter and major other air
pollutants like oxides of Nitrogen and some of the prevalent biological oxides. Each of the pollutants
has been briefly explained and its effect on various major cities around the world. For a wider analysis
on the Air Quality Index , the data set on Global air quality index is referred and various statistical
analysis are done using various tests focusing on the overall air quality index of a city with the
amount of Carbon Monoxide , Ozone and Particulate Matter 2.5 present in the air This study used a
combination of statistical analyses and machine learning models to examine the effects of ozone and
other pollutants on the Air Quality Index (AQI). The main conclusions show that ozone significantly
affects AQI, with models such as Random Forest and LSTM best illustrating this effect. While the
LSTM caught the temporal patterns of changes in air quality, the Random Forest model highlighted
ozone as a critical contributor among other contaminants. KNN offered information on regional
pollution trends, while Logistic Regression was a useful standard for classification work. These
conclusions were further supported by the statistical testing. Significant variations in AQI levels
according to pollutant concentrations were shown by ANOVA findings, and significant relationships
between pollutant thresholds and AQI categories were found by cross-tabulation studies. These
findings support the machine learning objectives and provide a thorough understanding of the ways
in which certain pollutants, especially ozone, impact air quality. There are two ways that this study
has ramifications. First of all, the results highlight how crucial it is to continuously monitor and
control ozone levels in order to enhance air quality. Second, the efficiency of fusing statistical analysis
with machine learning offers a solid framework for evaluating the effects of air pollution that can be
extended to additional contaminants and geographical areas. This research provides a great insight
into the effect of various air pollutants on the global air pollution trends. Several directions are
suggested for further investigation. Increasing the dataset's size to incorporate more areas and
contaminants may improve the results' generalizability. Deeper insights might be obtained by
incorporating more sophisticated machine learning algorithms and taking other environmental
aspects into account. Resolving the limitations that have been highlighted, such as possible biases in
the data and the requirement for more detailed temporal data, will enhance the study and make air
quality management plans more successful.
References
[1] Tran, V.V., Park, D., Lee, Y.C.: Indoor air pollution, related human diseases, and recent trends in
the control and improvement of indoor air quality. Int. J. Environ. Res. Public Health 17(8), 2927
(2020).
[2] Sathe, Y., Gupta, P., Bawase, M., Lamsal, L., Patadia, F., Thipse, S.: Surface and satellite
observations of air pollution in India during COVID-19 lockdown: Implication to air quality.
Sustain. Cities Soc. 66, 102688 (2021).
[3] Ekmekcioglu, D., Keskin, S.S.: Characterization of indoor air particulate matter in selected
elementary schools in Istanbul, Turkey. Indoor Built Environ. 16, 169 176 (2007).
[4] https://www.who.int/news-room/fact-sheets/detail/household-air-pollution-and-health
[5] Tiotiu, A.I., Novakova, P., Nedeva, D., Chong-Neto, H.J., Novakova, S., Steiropoulos, P., Kowal,
K.: Impact of air pollution on asthma outcomes. Int. J. Environ. Res. Public Health 17(17), 6212
(2020).
[6] Kumar, K.P., Unal, A., Jha Pillai, V., Murthy, H., Niranjanamurthy, M. (eds.): Data Engineering
and Data Science: Concepts and Applications. 2023.
[7] Madhurima, S., Mannooparambil, J.M., Kumar, K.P.: The Road to Reducing Vehicle CO2
Emissions: A Comprehensive Data Analysis. In: International Conference on Computational
Sciences and Sustainable Technologies, pp. 299 309. Springer, Cham (2023).
[8] Ashok, K., Reddy, P.A., Kumar,
Analysis. In: Kumar, K.P., Unal, A., Jha Pillai, V., Murthy, H., Niranjanamurthy, M. (eds.) Data
Engineering and Data Science: Concepts and Applications, pp. 335 349. Springer (2023).
[9] Satpute, B.S., Bharati, R., Rahane, W.P.: Predictive Modeling of Vehicle CO2 Emissions Using
Machine Learning Techniques: A Comprehensive Analysis of Automotive Attributes. In: 2023
3rd International Conference on Technological Advancements in Computational Sciences
(ICTACS), pp. 511 516. IEEE (2023).
[10] Sadriddinovich, J.T.: Capabilities of SPSS Software in High Volume Data Processing Testing. Am.
J. Public Diplomacy Int. Stud. 1(9), 82- (2023).
[11] Sicard, P., Agathokleous, E., Anenberg, S.C., De Marco, A., Paoletti, E. and Calatayud, V., 2023.
Trends in urban air pollution over the last two decades: A global perspective. Science of The
Total Environment, 858, p.160064.
[12] -based Precision Diagnosis of Lung
Sciences, vol. 76, no. 10, Oct. 2023, doi: 10.7546/crabs.2023.10.07.
[13] Yang, J. and Zhao, Y., 2023. Performance and application of air quality models on ozone
simulation in China A review. Atmospheric Environment, 293, p.119446.