78830 - Multivariate Statistical Methods for Credit Scoring

Course Unit Page

SDGs

This teaching activity contributes to the achievement of the Sustainable Development Goals of the UN 2030 Agenda.

Quality education Decent work and economic growth

Academic Year 2021/2022

Learning outcomes

At the end of the course, students know the basic concepts of statistical methodologies for taking business and financial decisions, and models for accounting and financial data for credit scoring.

Course contents

  • Introduction to statistical methods for credit scoring.
  • Recall of categorical variables: marginal and conditional independence, measures of association.
  • Logistic regression model: model specification, estimation and interpretation of model parameters, variable selection, goodness of fit.
  • Discriminant analysis: canonical discriminant analysis, model-based discriminant analysis.
  • Classification trees: CART and CHAID methods.
  • Methods for the estimation of the classification error and for the evaluation of the performance of the classifier.
  • Basics of neural networks.
  • Basics of latent variable models: latent class analysis.
  • Recall of cluster analysis for its use in credit scoring.
  • For each topic in the list, an analysis of case studies will be carried out by using the software R (at least 2 hours per week).

 

Readings/Bibliography

Compulsory reading

  • Elena Stanghellini (2009) "Introduzione ai metodi statistici per il credit scoring", Springer-Verlag. Available as Unibo e-book.
  • Teacher's lecture notes available on the platform Virtual Learning Environment at: virtuale.unibo.it

Suggested textbooks

  • Stefania Mignani, Angela Montanari (1997) "Appunti di analisi statistica multivariata", Esculapio (chap. 5 discriminant analysis, chap. 7 cluster analysis)
  • Sergio Zani, Andrea Cerioli (2007) "Analisi dei dati e data mining per le decisioni aziendali", Giuffrè Editore (chap. 8 distances and similarity indexes, chap. 9 cluster analysis, chap. 11 classification trees, chap. 12 neural networks)

Teaching methods

Theoretical lessons and practical activities in computer laboratory with R (individual or group work).

If online only or blended lessons will be done, students should work with their personal laptop by installing the R software, see: https://www.r-project.org/

Lessons are not recorded.

Attending the lessons is not mandatory but it is strongly recommended.

As concerns the teaching methods of this course unit, all students must attend Module 1, 2 on Health and Safety online.

Assessment methods

Final written test to assess the knowledge of the statistical methods both from a methodological and from a practical point of view.

The written test consists of open-ended items dealing with both the statistical theory and the interpretation of outputs produced with the software R. In the teaching material it is possible to find an example of written test. The test duration is from 90 to 120 minutes. During the test, it is possible to use a pocket calculator only.

For each item, it is given a score. The sum of the scores is equal to 32. The final mark is expressed on a scale of 30 and it is calculated by the sum of the scores obtained in the items. The "lode" is given to students who have a total score equal to 32.

The final mark corresponds to the following description of the overall achievement level reached:

< 18: not sufficient (exam failed)

18-23: sufficient

24-25: satisfactory

26-28: good

29-30: very good

30 e lode (30 cum laude): excellent

Prerequisites: mathematics, statistics, probability, inference.

Attendance of the lessons is strongly recommended.

Students enrolled in the academic year 2016/2017 or before can choose if taking the written or the oral exam. Please write an email to get information.

The methods of carrying out the test will be those provided by the University (in the presence, online or both). However, the way of carrying out the test does NOT change the type of exam that is WRITTEN. In the case of online examination, the test will be managed with the EOL (On-Line Exams) applications for reception/delivery and Zoom for surveillance. Further information will be sent by email to the list of registered students.

 

Teaching tools

Slides and lab material, software R.

Office hours

See the website of Mariagiulia Matteucci