Estimation Of Credit Limit For Microcredit

Microcredit is short-term, small-scale, frequent and urgent in its character. Currently, microcredits are mostly credited by credit managers and credit committee case by case in banks, which is slow and costly, also lacking of unified rules. Way to solve the problem is to replace the current business style by introducing mathematics methods into the estimation process of credit limit for microcredit, making the process informative and standardized. By fitting the behavior patterns of loan applicants into mathematics models, it is more quick, precise and less costly for banks to make decisions on the next case.Two kinds of models were used to make a study on the estimation of credit limit for microcredit based on the physical data, which were Ordinal Logistic Regression model and Fussy C-means Clustering algorithm. Empirical analysis showed that Ordinal Logistic Regression model was better than Fussy C-means Cluster in fitting with the physical data used in this article. As a deductive method, the regression model plays good in imitating the experts in evaluating the applicants' ability to pay back the loan. While the fussy cluster model is unsupervised, which performed not as well as the regression model when based on the data coming from the prophase of the credit process. Besides, Ordinal Logistic Regression model is rigorous and forecasts precisely and easy to build, which makes it a good solution for credit limit estimation.While the model came up in the article helps in improving its efficiency, the estimation of credit limits for microcredits still remains a problem because of the quality of raw data was far from satisfactory. Too much noise in the data constrains the actual result of the model. Standardization in data collection and finding ways to reduce noise means a lot, which was also discussed in the last part of the article.