Logistic Regression: Credit scoring in microfinance and banking: 3

Credit scoring has much to offer to microfinance institutions and smaller banks. It is a simple and powerful tool in reducing operational costs and loan losses. Moreover, it speeds up the loan request evaluation process. Reducing costs and improving customer satisfaction? Who would not be interested? In this series of videos, we will discuss four simple credit scoring techniques: expert approach, Bayesian scoring, logistic regression, and the Altman Z score. Our goal is to demystify the mathematics and logic behind these models. We will explain in plain terms the underlying principles. Using Excel simple models, we show how to use these techniques in practice. Even smaller institutions without sophisticated IT-systems can put in place these scoring techniques. This third video in our series covers logistic regression. In the first part of the video we discuss the linear regression and in the second part the logistic regression.
Good luck!
André Koch Stachanov Solutions & Services
Tags:
Credit scoring,Logistic regression,Excel,Stachanov,André Koch, Stachanov Solutions & services,s-curve,credit rating,=slope(),=intercept(),microfinance,excel modelling,risk management,risk model,risk modelling,regression,regression analysis,multilinear regression,forecasting,predictive model,credit risk analysis

Пікірлер: 18

  • @hifi.880
    @hifi.8803 жыл бұрын

    grate initiative thank you

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    3 жыл бұрын

    You are welcome!

  • @dawlenceable
    @dawlenceable3 жыл бұрын

    Thankyou!! Very useful video. Just one question, how did you determine the coefficients for various attributes such as agriculture, age bracket etc.

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    3 жыл бұрын

    Hahaha, yes, this is a good question. As I mentioned in the video this is a beyond the scope of my presentation. I just wanted to explain the basics of logistic regression with a simple example. Although, calculating these a ratios is somewhat more difficult, it is entirely possible to do this in Excel. What you need is enough historical data. Say, a thousand loan records with age, marital status, etcetera and whether or not the loan was reimbursed. Now, using your function you need to tweak the coefficient values and the constant in such a way that your function generates in as many cases as possible a match with the historical data. So, when the actual loan was not reimbursed, also you function should predict the same. You can do this using Excel Solver. I made another video on optimising a loan portfolio along the axes of risk and return. This is the technique you would use: kzread.info/dash/bejne/pKF2mc2noqbQZJs.html I hope this answer points you into the right direction. Best, André Koch

  • @dawlenceable

    @dawlenceable

    3 жыл бұрын

    @@StachanovSolutionsServices Wow thanks man!! Wasn't expecting such a comprehensive reply tbh. Love all your videos! keep up the good work! Cheers

  • @ronaldpereira4436
    @ronaldpereira4436 Жыл бұрын

    Hi, just watched the video, many thanks for the explicit explaination. Understood the logistic regression and PD clearly. just trying my luck here, have you made any videos on how to convert the logistic regression to a score card for application score card. Any help would be much appreicated

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    Жыл бұрын

    Dear Robert, thanks for you compliments and for your question. No, I have not made a video on mapping this to a scorecard. Sorry, best, André

  • @ronaldpereira4436

    @ronaldpereira4436

    Жыл бұрын

    @@StachanovSolutionsServices , Thank you for your kind revert. Would you consider making one any time soon or guide me where I can look this up where it is explained in a lay mans terms.I am looking for a broad understanding to begin with and then work my way. Thanks Ronald

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    Жыл бұрын

    @@ronaldpereira4436 Sorry, to disappoint you, but I am quite busy and, frankly, this is not at the top of my list as I am mainly working in micro-finance. You have to do some research yourself. All the best to you ! André

  • @varunrao2721
    @varunrao2721 Жыл бұрын

    Hi thanks for the video. Could you please let me know in the multi linear regression formula (sector + gender + age + constant) from where did you get the constant value which was 0.49?

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    Жыл бұрын

    Dear Varun Rao, sorry for the delay in answering you. The video was meant to give an idea of the principles behind logistic regression and indeed does not provide a full cookbook on how to do this in a more complex linear function with multiple slopes. In brief, to calculate the slopes as well as the constant you need to perform a mathematical optimization. You could do this using Solver. If you like, we can send you the spreadsheet. Please, contact me at andre@stachanovcom All the best, André Koch

  • @varunrao2721

    @varunrao2721

    Жыл бұрын

    @@StachanovSolutionsServices Thank you Andre!

  • @swetapatra
    @swetapatra3 жыл бұрын

    how did we calculate the coefficients in 8:09?

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    3 жыл бұрын

    Hello see my answers below to Dracarys . In brief, they are obtained in an optimisation process. How to do this is beyond the scope of this video. Best, André

  • @nik6920
    @nik69203 жыл бұрын

    Where to take & see all the common risk factors, their parameters and how all of them tied in a formula that returns approval result for credit?

  • @StachanovSolutionsServices

    @StachanovSolutionsServices

    3 жыл бұрын

    Hello, I am not sure how to interpret your question. I think you refer to the coefficients that are used n the logistic regression function. There are calculated in an optimisation process. How to do this was beyond the scope of this introductory video. Best, André Koch