Using historical data and statistical or machine learning models to forecast future outcomes or estimate the likelihood of events.
Predictive analytics goes beyond describing what happened to estimating what is likely to happen. It uses patterns in historical data to build models that can be applied to new data. Common techniques include regression analysis, decision trees, and machine learning algorithms.
Predictions are probabilistic — they express likelihood, not certainty. A good predictive model communicates its confidence level and the conditions under which it is reliable.
Example: A transit authority uses predictive analytics to forecast ridership on each route for the next week, based on historical ridership patterns, weather forecasts, and upcoming events. The predictions help schedule the right number of buses to meet expected demand without over-deploying resources.
Predictive analytics requires careful attention to data quality and model bias. A model trained on biased historical data will produce biased predictions.