Forest fire is one of the most dangerous natural hazards of planet earth now. Here presenting an approach where it has been trying to analyze the danger of fire for the forests of Brazil. The dataset contains data from 1998 to 2017 for all the states of Brazil where a forest fire has been caught throughout the year. An unsupervised approach like—K-Means clustering, Fuzzy C-Means, and Apriori algorithm was used to do so. This is a large dataset containing 6454 unlabeled data, to build a model with them K-Means clustering seems helpful. It tries to build subgroups (clusters) of similar data points from a large group. Fuzzy C-Means is also an unsupervised algorithm and it’s working process is similar to K-Means clustering. By using K-Means clustering, Fuzzy C-Means, and Apriori method the regions which are in risk of fire danger were detected.