ANNOUNCEMENTS
In India forest fire Events has been in rise since the last 5-6 years, For the state of Odisha a rise has been seen for the last 6 years (FSI). A study has been conducted for the state of Odisha so that the rise in forest fire events can be known and a trend can be seen and it can be cross check with the forest fire season cycle of the Odisha February-July (FSI) and also study the policy framework for forest fire and should provide recommendations on the policy grounds. As part of the methodology initially have collected the parameters that are responsible for the forest fire (human factors are not included in it) and then were used for the Training and testing of the ML Model with 87% and 82%. As the accuracy of training and testing data sets are near this shows that our ML model will be the best for our data and further future work like developing prediction model for time series analysis can be done easily. As for the policy study have gone through the National and state policy and looked for the gaps that exists and have suggested the Recommendations like to include human interference whenever doing the analysis for the forest fire as it is the major cause of forest fire followed by NTFPs. During the study have found that the forest fire events have been increased drastically in Odisha (FSI) and an increasing trend can be seen here the data we get is from SNPP VIIRS that will account for large and small forest fire. So, from the study we get to know that the currently used prediction model I not much accurate for forest fire prediction as have only 70% accuracy (ISFR,2017). We have Used Random Forest classifier with 82% accuracy in testing so it can be used in prediction anal analysis further also for part of policy recommendation India should have the seasonal weather forecast system for the early prediction of the forest fire and in that our ML model can be useful.
Keywords: Forest fire, ML Model, NTFPs, Policies , Recommendations.
