Summary: | 碩士 === 國立中興大學 === 應用數學系所 === 107 === From previous fire statistics, the database excluded the case of weed fire or false alarm fire due to minor loss caused from fire incident. The actual number of firefighting attendances have larger numbers than the database. The past statistics analysis uses the record from database results the difference from actual practice. In this paper the actual number of firefighting attendances are used as analyzed data to reveal the reality.
The data of this paper uses fire statistics from September 2016 to August 2017 in Taichung City. Find possible factors that cause fire occurrence through literature review and descriptive statistics and use two-way analysis of variance (ANOVA) method to assess. The analyzed results from ANOVA is to establish Poisson Regression model for better understanding the connection between fire factor and fire incident.
This paper find that the overall fire incidents affected by seasons, time, and regions. Building fires have more impacts on building floor and materials used rather than time and regions. Apart from seasons, time and regions influence, weather factors include temperature, humidity and rainfall also affect weed fires.
|