The course is split up into 4 online interactive training sessions each 3.5 hours. For detailed content refer Course Outline.
Date : 30th November – 3rd December, 2020
Time : 13:00-17:30 CET
Price per participant: 1,100 EUR *
*Alumni discount is 10%
*Group discount is 10%
This course is of interest for finance professionals in banking, leasing, factoring, consumer and small business lending, fintech, telecoms and health care. The content is particularly relevant for credit managers, risk analysts, business process engineers, IT specialists, auditors, consultants and many others interested in predictive statistical modelling.
Today, big data and scoring are driving real-time business decisions everywhere, not only in finance. The sample data sets and practical case studies in this course will focus on credit risk assessments in granular consumer and micro/small business portfolios. The immediate objective is to build better application and behavioral credit scoring models that can support or automate credit granting decisions. However, once you start analyzing credit data sets and working through the scoring process, the versatility of the methodology becomes immediately apparent: instead of predicting default behavior, one can just as easily score the probability of fraud, predict the client’s next purchase, target low risk drivers for special car insurance offers etc. The only pre-requisite is a university-level analytical perspective and the willingness to engage with fundamental statistical concepts. Knowledge of R, SPSS, or other statistics packages is helpful, but not required. We will demonstrate all calculations using standard Excel functionality and inexpensive add-ins such as XLStat.com or open source tools such as Jamovi.org.
Lecture, discussion, individual and group exercises, case studies and reading materials
Unit 1 – The Context of Statistical Credit Scoring
All the buzzwords explained: credit bureau scoring, expert scoring, credit rating, application scoring, behavioral scoring, scoring on alternative data: mobile, social media and psychometric observations. The ethics of scoring. How to address data privacy and data confidentiality concerns.
Unit 2 – The Statistical Scoring Process
Defining the analytical target. Setting the good/bad variable. Selecting the relevant population of observations in terms of time window, sectors, geographies. Generating and cleaning up the data set. Dealing with outliers and missing values. Coding, discretizing / transforming / recombining the candidate variables. Estimating a model using various algorithms. Validation testing on the reserved sample.
Unit 3 – Recalibration and Maintenance
Test and recalibrate existing models periodically for continued performance. Integrating new seasoned observations and additional data points. Making forward looking expert adjustments in order to capture external effects, such as the medium-term economic impact of the COVID-19 pandemic.
Unit 4 – Outlook: What else can we score?
Practical example of deploying scoring for fraud prevention. Plus additional use cases for scoring using ideas contributed by participants in the course.