Application of Data Mining Approach for Profiling Fire Incidents Reports of Bureau of Fire and Protection

The study aims to create a fire model that can generate pattern to obtain profile of fire incidents in Manila that can help facilitate the assessment of fire incidents by providing a quick, thorough, and scientific analysis of fire data, the outputs of which in turn can be utilized as basis for the formulation of new programs and fire prevention activities. The researchers applied K-means clustering approach and Elbow method using Python scikit-learn tool to process the fire incidents dataset and produce the number of clusters with centroid and correlation heatmap which resulted to five clusters. The most noticeable result is the "Day" attribute, Wednesday and Thursday, which are similar across all five clusters. In addition, most fire incidents occurred between 12:00 NN to 4:00 PM, around lunchtime and afternoon siesta. The result also shows that the poor are the most vulnerable likewise, the middle class. It concludes that government can reconsider its services on fire protection issues and fire disaster management. Data mining techniques applied to fire incident reports was limited in the Philippines. With this. more work should be carried out using prediction which can help in predicting the presence of a fire incident so that firefighter could immediately respond.