Abstract: Guizhou is rich in forest resources and has a high forest coverage rate, and the interlacing of forest land, farmland and villages with natural and social factors is superimposed, making it the hardest hit area of forest fires in China. Therefore, it is of great significance to analyze the driving factors of forest fires in Guizhou province and carry out fire risk zoning to improve the fire prevention and control ability in this region. In this paper, the forest fires in Guizhou province in the first half of 2024 are analyzed by considering the meteorological, topographic and human factors. Based on four different machine learning models,including random forest, convolutional neural network, extreme learning machine and adaptive lifting algorithm, the grid search method and Bayesian optimization method were used to optimize the parameters to construct the optimal prediction model,and the forest fire risk distribution map of Guizhou province was drawn according to the prediction results. The results show that the prediction accuracy of the four models is more than 85%, and the accuracy of the random forest model reaches 91%.Meteorological factors such as daily maximum temperature, total daily precipitation and average wind speed are the key factors affecting the occurrence of forest fires in Guizhou province. In addition, parts of Guizhou province, such as Guiyang, Bijie, and Qiannan Buyi and Miao Autonomous Prefecture, are at higher risk of forest fires in July and August 2024. This paper provides reference information for the prevention and control of forest fires in Guizhou Province, which is helpful for relevant departments to formulate more effective fire management strategies.