The Customer Missing Analysis Of Kunming Telecom

The article use some technology such as Data Warehouse, statistical analysis, Data Mining, etc. and deeply and sophisticated research the subject of the PAS customer missing of Kunming telecom. The author has surveyed and studied in Kunming telecom almost one and a half years, and accumulate much experience, theory and field knowledge.Aimed at the PAS customer missing, the article takes some customer missing information which occurred in 3 months of 2005 as sample. And according to fact of the Kunming telecom's ODS data, the article made certain of basing characterized vector and target variable, individual sample influenced missing. The acquiring and preprocess of sample data is a very important work to data mining. The article has spend much of time in research of relatively technology and statistical method, at last we decide a project which use OLAP, factor analysis, interaction analysis, relativity analysis, several judgment analysis. This including preprocess such as acquiring characterized vector of sample space, influence, fixing and forecasting. Then, the article offers a good sample set, and builds a Data Warehouse based on the subject of PAS customer missing. Based on build a "wide table", we use clustering method to classify the characterized vector, determine the distribute characteristic of many character weight such as custom's value zone, natural attribute, terra zone. And sum up much commonness, and combine field experience, obtain decision trees create condition, and guide the building of decision trees.The practice proved, the technology of the article used is feasible, the mid result take some effect to missing analysis of Kunming telecom, the knowledge we discovered has some rationality and reference value.