This course is a CrashProgram (short course) introducing the concept of a structured review of risk data.
Step by step we build the knowledge required to review the suitability of data for a given purpose and how to report the findings.
Object-oriented programming and techniques (OOP) such as using classes and inheritance are common in many application programming environments but don't travel well outside computer memory. When considering data science tasks and objectives the transition from object hierarchies to data structures (and vice versa) is not always straightforward. In this short course we explore how some programming languages, data formats, database API's and web frameworks handle hierarchical classes.
In this short course we explore how some programming languages, data formats, database API’s and web frameworks handle hierarchical classes.
Object-oriented programming and techniques (OOP) such as using classes and inheritance are common in many application programming environments but alas don’t “travel well” outside computer memory. The potentially intricate relationships of objects (both the data they hold and the meaning and possible uses of the data) are not easy to transfer (except of-course by full replication of code and data). Hence when considering data science tasks and objectives that involving exchange of data, the transition from object hierarchies that live inside memory, to data structures that can be exchanged with another computer is not straightforward.
The Open Risk Academy course NPL270672 is a CrashCourse introducing the EBA NPL Templates.
We start with the motivation for the templates and the domain of credit data (to which NPL data belongs). We discuss three core classes that capture the essence of lending operations from a lenders point of view (Counterparty, Loan, Collateral). Next we explore classes that capture events in the lending relationship lifecycle (which we term NPL Scenarios). We look into the main data types: elementary data types, choice lists, arrays and unstructured text. We close with discussing some more complex issues involving graph and timeseries data.