Energy Credit Best Practices – Chapter: Information Technology http://ccro.org © Copyright 2022, CCRO. All rights reserved. 28 able to generate historical views is helpful so that decisions can be explained in terms of what data was available when the decision was made. 3.4.5 Version control Version control is a process designed to keep track of multiple versions of software. Any system that provides change tracking and control over software and documentation can be considered a version control process. The practice has been a part of software development or printed material almost as long as writing has existed. The purpose of version control is to ensure that the software's content changes are documented, and the effects open to other components within the Credit Information Ecosystem are understood. At the same time, version control is often carried out by a separate application, word processors, and spreadsheets. Version control allows servers in multiple locations to run different versions on different sites, even while those versions are being updated simultaneously. 3.5 Advanced Information Technology Customarily, IT strategies have revolved around the “enablement” of business roles such as Credit Risk managers, analysts, and risk professionals to perform their day-to-day tasks. Today, advanced IT strategies focus on growing data maturity, which ultimately leads to a higher level of confidence in reports and analytics sourced from the core datasets. The following Advanced IT concepts provide unique insights and opportunities affecting Credit Risk management. 3.5.1 Artificial Intelligence (AI) Often, AI is associated with sci-fi movies and is more recently drawing calls for increased regulation to limit perceived threats. However, when applied appropriately, AI can facilitate more benign Credit Risk functions, including evaluation, planning, calculations, analytics, and decision- making. 3.5.2 Machine Learning (ML) This is a subset of Artificial Intelligence and contains a baseline or starting point in mathematics and statistical algorithms. Machine learning depends on learning over time and typically performs best when provided with large, diverse amounts of data from which to “learn”. ML involves the following three areas: • Predictive Analysis – As the term “predictive” describes, these models are used to provide insight into future events, patterns, and trends. A typical business implementation of Predictive Analytics • involves decision trees where the system “learns” from past decisions and outcomes and uses it to “predict” the future path for success.