Improve Business Decision

Many business models are based on the accuracy of decision-making. Loan applications, warranty repair approvals, retail pricing, and wind turbine placement to name just a few. In these situations machine-based decision models are often preferred as human judgment can often be subject to bias and unreliability.

Another beneficial factor of machine-based decision-making is speed and scale. A machine can make many more decisions than a human but comes with the risk that quality has to be monitored to ensure the model stays up to date and can reliably defer decisions to a human review when required.

Using Machine Learning to automate the assessment of creditworthiness for loan applicants at a bank.

Decision models are often framed as multi-class classification problems for machine learning algorithms. Using historical data the built model is normally tested using cross validation for robustness. Cross validation shuffles the training data into different train and test splits to ensure as far as possible that the model is robust to unseen data.

A large Multi-national bank used C3.ai, an Enterprise AI platform, to quickly assess the creditworthiness of its customers for lending decisions. Using 9 different data sources the bank was able to achieve a 30% reduction in average cycle time resulting in a $100m expected annual revenue increase2.