Pattern Extraction and Rule Generation of Forest Fire using Sliding Window Technique

Abstract

The sliding window technique is being used to extract patterns of forest fire which consists of burnt area size, temperature, relative humidity, wind speed and rainfall. The initial data is being transformed by changing the continuous values of the attributes into categorical value. Extracted patterns are then being grouped based on the size of burnt are. Rules are then generated by transforming the categorical values into intervals and the merging different records into the same rules. The rule generation stage produces eight distinct patterns of meteorological conditions that could predict the size of forest fire.