<![CDATA[The events of forest fires can occur naturally or artificial that impact environmental damage and loss of all aspects. Indonesia's wildfires are increasing annually. This is because Indonesia has many peatlands and rainfall in the dry season is less than half the normal rainfall or known as the El Nino Southern Oscillation (ENSO) phenomenon. Early indications forest fires can be known through a fire point (hotspot). In the years 2010 to 2018 found 14.070 fire points in Java region. One way to detect land fires is to divide the data of the fire points into groups using the Self Organizing Map (SOM) method. To measure the quality of the formed cluster, the Silhouette Coefficient (SC) algorithm is used. Based on the test results obtained the highest SC value of 0.248945455 with the value of neuron count is 3, alpha value is 0.1, maximum epoch value is 18 and the value of reduction of learning rate is 0.1. In 2017 the resulting SC value was 0,23416068940874324. The result is that East Java region has a big chance of land fires if seen from the point of fire that appears and confidence value.]]>
Copyrights © 2019
