Abstract:
Clean air is crucial for animal life as well as human health since it is associated with
a number of deadly illnesses, including cancer. However, due to the world's rapid
urbanization and population growth, activities including housing, industries, ships,
and farming contribute to air pollution. As a result, pollutants in the air have become
a severe problem in many cities, especially in developing countries like Bangladesh.
Maintaining indoor air quality requires frequent monitoring and forecasting of air
pollution. As such, ML has demonstrated potential in predicting the air quality index
(AQI) more accurately than conventional methods. An indicator of the condition of
the air is the index for air quality (AQI). It computes the short-term impact of
moderate exposure on an individual's health. The AQI's mission is to raise public
awareness of the harmful effects that nearby contaminants have on health. The
quantity of pollutants in the environment has significantly increased in Indian cities.
By using the AQI for Bangladesh's capital, Dhaka, we are focusing on a few variables,
starting with PM2.5 in 2017 and going up to 2022. The goal of the study is to find out
how successfully NLP techniques identify and classify activity in AQI categories.
Using labeled data, controlled instruction teaches an algorithm how to accurately
forecast outcomes and classify AQI data. For this purpose, a variety of machine
learning models were employed, including XG Boost, a Random Forest, K-Nearest
Neighbors, Naive Bayes, and Linear Regression. Following data analysis, the most
accurate classifier, with a 99.81% classification accuracy, was the Random Forest
classifier for AQI values that fell into six categories: Hazardous, Unhealthy, Very
Unhealthy, Good, Moderate, and Unhealthy for Sensitive Groups. To produce a web
prototype, the AQI category is finally classified using a Random Forest model.
