在天然氣輸差分析的研究中，輸差原因錯綜復雜，但在其影響因素數據中許多潛在有價值的規律未被發現。在各種機器學習算法中,決策樹以其簡單容易實現等特點被認可。本文首先介紹了分類器的基本概念和決策樹構建思路，然后講述了在天然氣信息數據庫的基礎上如何建立決策樹分類器，接著利用xml 的優越性進行存儲，并依據創建
的決策樹對數差進行預測。
關鍵詞 天然氣數據 輸差分析 數據挖掘 決策樹 C4.5 Xml
Abstract In the studies of analysis on measuring error of natural gas ,reason of measuring error is complex, but the potential and valuable rules hid in the result data have not been discovered. The algorithm of decision trees is well known due to simpleness and easy to realize in machine learning.Firstly we describe how to construct a decision tree classifier on the information database made by natural gas corporation,then store it taking advantage of xml,and forecast measuring error by the desion tree created.
Keywords natural gas data analysis on measuring error Data mining Decision tree C4.5 Xml