Learning objectives
- Define regression in practical language
- Distinguish regression from classification
- Identify common real-world regression tasks
Introduction
Regression is a supervised learning task where the target is a continuous numerical value rather than a class. Instead of predicting spam or not spam, a regression model predicts a number such as next month’s sales, house price, patient wait time, or electricity usage.
Regression is valuable because many operational and financial decisions depend on estimating quantities. A business needs forecasts, a school may need resource planning, and a logistics company needs travel time estimates.
The main difference from classification is simple but important: regression predicts how much, while classification predicts which category.
What makes a problem regression
If the output can take many numerical values along a scale, the task is likely regression. Examples include revenue, risk score, energy consumption, and test score prediction.
Some tasks can be framed either way. For instance, predicting exact house price is regression, while predicting whether a house is luxury or standard is classification.
How regression outputs are used
A regression output may feed directly into planning, budgeting, staffing, or pricing. Because the outputs are numerical, it is often useful to examine error size rather than simply whether the prediction was correct.
This also means outliers and unusual cases matter. A small average error may still hide large mistakes on important records.
Interpretation and caution
Regression models can sometimes be easier to interpret than more complex systems, especially when the relationship between features and target is relatively clear.
However, they still depend on good data, stable patterns, and thoughtful evaluation. A model trained on old market conditions may fail badly when the environment changes.
Examples
House price estimation
A model predicts the likely sale price of a property based on features such as location, size, age, and amenities.
Demand forecasting
A retailer estimates how many units of a product will be sold next week in each region.
Travel time prediction
A ride-hailing service predicts the number of minutes required to reach a destination based on traffic, distance, and time of day.
Exercises
- Explain regression in one sentence for a beginner audience.
- Turn three business questions into regression problems.
- What is the difference between predicting a category and predicting a number?
- Why might changing market conditions damage a regression model?
- Give an example of a task that could be framed as either regression or classification.
Key takeaway
Regression is the right tool when the goal is to estimate a meaningful number rather than choose from a set of categories.