Smart Detection of Passive Smokers using Machine Learning

• Published in: International Conference on Computing Communication Control and Automation (Online) pp. 1 - 6
• Main Authors: Shah, Pratham, Parekh, Astha, Bajaj, Divyansh, Swain, Debabrata, Kumar, Manish, Bhilare, Amol
• Format: Conference Proceeding
• Language: English
• Published: IEEE 23.08.2024
• Subjects: Accuracy; chi-square based feature selection; Diseases; Encoding; Feature extraction; GridSearchCV; hyperparameter tuning; machine learning; Null value; Passive smokers; Radio frequency; Random forests; Smoking detection; SMOTE; Supervised learning; Training; Tuning
• ISSN: 2771-1358
• DOI: 10.1109/ICCUBEA61740.2024.10774944

Smoking is a destructive and addictive practice of inhaling burning plant material. It produces nicotine in the bloodstream that creates a negative impact on the bones, hormones, DNA, eyes, and the immune system. This has become the source of a great concern, especially amongst the young generation. Machine Learning is used to find a pattern of biological attributes to identify possible smokers. The process of teaching a machine to do intelligent classification from a sizable archive of historical data is known as machine learning. The aim of this paper is to develop a learning model that can detect passive smokers through features like eyesight, weight, fasting blood sugar, cholesterol and more. On this dataset, a number of data preprocessing operations are carried out to build a reliable model which include checking for null values , encoding the data , balancing the dataset and scaling it and selection of highly correlated features. A variety of supervised learning models are employed in the model training process to create a robust machine learning model. From the set of models applied, K-Nearest Neighbour (KNN) (84 %) and Random Forest (RF) (86 %) gave the highest accuracy. However, implementing Random Forest with Hyperparameter tuning using GridSearchCV resulted in a higher accuracy (87 %).