Modeling mechanisms of lending to individuals on the basis classification algorithm on an example of microfinance company

<p>The topic graduate work is Modeling mechanisms of lending to individuals on the basis of the classification algorithm on an example of a microfinance organization.</p><p>Volume of graduate work is 57 pages, including 14 figures and 16 tables, but not including appendices.</p><p>The aim is to construct an objective and reliable methods of evaluating the reliability of the borrower microfinance company using classification algorithms</p><p>Thus, the important results were obtained in the course of the research.</p><p>In the first chapter was studied the available publications available about scoring, its application and construction techniques. Under construction techniques we understand statistical methods and classification algorithms, which do not yield to each other, and can be used either together or separately, depending on the policy of the organization, using a scoring model.</p><p>In the second chapter we have discussed in more detail known modeling approaches grading system that allow us to determine the &quot;quality&quot; of potential borrowers, the mechanism of the evaluation - the intermediate stage between the input data (characteristics of the borrower) and the dependent variable (the status of the loan).</p><p>In the third chapter has described the whole mechanism of collecting and processing information for further research on the division of the population learning and test sample on which the experiments were performed on the applicability of different methods of classification problems and testing the results obtained, respectively.</p><p>For statistical and mathematical calculations were applied methods discussed in Chapter 2.</p><p>As a result of the current study, we obtained acceptable results for each of the proposed methods, from which we can conclude that the available methods for solving classification problems for the credit industry and other industries can be applied to microfinance organization. Besides using logit and probit regression coefficients were obtained individual estimates for the characteristics of the borrower, which assess the impact of each performance than can benefit investigated company.</p><p>Each classification method yields the individual results of the evaluation, but they give a certain percentage of errors in the estimation of a &quot;good&quot; bad customers and vice versa. This can be seen in the comparative tables of actual and predicted values ​​for the variable &laquo;loan_status&raquo; and graphs. Of course, the error rate depends on the probability of establishing a threshold above which investigated the microfinance company receives a potential borrower, which will return the loan, this does not exclude the loss of &quot;good&quot; customers.</p><p>Conclusion contains the main conclusions of the study and recommendations for further possible applications of credit scoring systems.</p>