The Application Of Decision Tree Building Individual Housing Loan Risk Assessment Model

The paper analyzes large data of individual housing loan by decision tree technique of data mining. Through uncovering the hiding models in the large data, this paper finally gets a risk assessment model for individual housing loan. These studies mainly include data selection, data preparation and constructing model by data mining tools. These three parts are spread out based on some relevant theories of data mining and decision tree, according to business need and the general steps of data mining.Facing the objective conditions of lowing credit assets quality and enlarging risk in our country's commercial bank, drawing foreign banks' successful experience of building customer credit appraising systematic by decision tree technique of data mining, under the premise of clear mining purpose, we go deep into the concepts of data mining, decision tree , data collect , data preparation , clustering analyse and the model evaluation. Combining with the main problem of this paper studying and the characteristic of these data, analyzing and comparing various methods, we adopt the suitable data collect and preparation methods to deal with the half of data in the data set. Thus we get a new data set that is appropriate for mining. And then we apply C4.5 Algorithm of decision tree and K-average Algorithm of clustering to mine and analyze these data. Finally we structure a risk assessment model for individual housing loan. Using another half of the data as the testing sample, we find this model have stronger forecast ability. It has been the best model that can be adopted by the commercial bank at present and it is worth being extended in practice.To speak concretely, contents that the thesis studies can be shown in the following aspects:1 . Aiming at the phenomenon of the rising individual housing loan infringement rate in our country and referring the achievement that the abroad banks, ??the paper proposes that using decision tree technique of data mining to construct risk assessment model for individual housing loan will bring much direct beneficial result to commercial bank.2. By comparing with classify algorithm, decision tree technique and classic C4.5 Algorithm are chosen to study this problem of risk assessment for individual housing loan. And then, the paper introduces the theories of decision tree and C4.5 Algorithm.3. Based on the process and purpose for data mining, the paper analyzes large data in the primitive database. Random array generator takes randomly out half of the data as the sample data to get a new data set for individual housing loan. This data set is the basic of data mining. Another half of data are used for testing.4. Considering the quality of data and the calculating efficiency, this paper studies and applies some important data preparation methods and selects appropriate data preparation methods to deal with the data set of individual housing loan. And then we get a new data set that is appropriate for mining.5. Based on theories of data mining and decision tree, we use decision tree classify technology, clustering technology, C4.5 Algorithm of decision tree and K-average Algorithm of clustering to get a risk assessment model for individual housing loan and some regulations.6. We use the other half data in primitive data base to test and evaluate the risk assessment model for individual housing loan. We find that the model has higher predictive accuracy. Finally, this model can really be used to help bank loan department analyze every loan and help bank leader make correct decision.