Predict binary outcomes: Building decision support models with Excel

Binary outcome modelling is used across many fields where decisions depend on the likelihood of an event occurring. Typical applications include quality control, finance and credit decisions, operations and reliability, marketing and customer behaviour, human resources, and health and policy analysis. In these contexts, the goal is not only to predict an outcome, but to estimate and interpret the probability of success or failure to support a decision.

In this course participants will learn how models to predict binary outcomes work, and using the most common approach, the logistic regression, will learn how to develop a model to predict the probability of observing a YES or NO outcome, creating a decision boundary to determine success or failure, and how to interpret these predictions in a practical context.

The emphasis is not on mathematical detail, but on the practical aspects of building the model, checking its performance (hit or miss), and decide whether it is useful for real‑world decisions.

Through guided, hands on exercises in Excel, participants will:

• propose models for binary outcomes
• estimate and interpret probabilities
• evaluate how well a model separates different outcomes
• communicate results, limitations, and uncertainty clearly

By the end of the course, participants will be able to use binary outcome models as decision‑support tools.

Keywords: Binary outcomes, probability prediction, decision support, logistic regression, classification performance, confusion matrix, ROC curve, Excel‑based analysis.

Part I: Variables with two possible outcomes (binary)
Recognising situations where outcomes have two possible results
Evaluating how variables relate using tables
A review of probabilities and odds

Part II: proposing the model 
Proposing a model to estimate the probability of an outcome
Estimating model parameters using Excel
Interpreting model outputs in practical terms
Understanding how predictors influence the likelihood of outcomes

Part III: Evaluating the model 

Evaluating model performance via performance analysis: hit matrix, performance ratios, precision, accuracy.
Evaluating model capabilities from a statistical perspective: deviance and comparing against a reference null model

Identify situations where a binary outcome model is appropriate
Construct models in Excel to estimate the probability of binary outcomes
Interpret predicted probabilities in a decision‑making context
Evaluate model performance using appropriate classification summaries such as the hit or confusion matrix, the ROC curve and other statistical metrics.

Classroom delivery coupled with hands-on applications via case examples using regression template files and interactive sessions.