A Comparative Analysis of Machine Learning Models for Fire Alarm Systems

The current study compares the efficacy and accuracy of many machine learning models using the Sensor-Fusion Smoke Detection Classification dataset for minimizing false alarms. The fire alarm system's accuracy was predicted using the LightBGM Algorithm, Naive Bayes, K-Nearest Neighbors, Logistic Regression, Random Forest, Support Vector Machines, Decision Tree classifier, Gradient Boosting, and AdaBoost classifier. Confusion matrix was employed as the assessment metric, and test and training was utilized to guarantee the models resilience. The findings indicate that the LightBGM algorithm achieved the maximum accuracy of 99.9%, ahead of Decision Tree and K-nearest neighbor, which had respective accuracy of 99.6% and 99%. With 73.7% accuracy, Naive Bayes achieved the lowest accuracy. These findings show that the LightBGM algorithm is the most suitable for estimating a fire alarm systems accuracy. The findings of this study could help in the creation of fire alarm system prediction models that are more precise.