4 Regression Free Online FlashCards

What does regression predict?

Regression is a supervised learning task that predicts a numerical target from one or more input features.

In a regression dataset, what are features and the target?

Features are the input information, such as floor area or bedroom count; the target is the numerical value to predict, such as sale price.

How does a mean baseline make predictions?

A mean baseline predicts the average target observed in the training data for every example. It provides a simple reference for judging whether a model is useful.

What does the area coefficient mean in the rent model?

Linear regression predicts a target as a weighted sum of features. In the rent example, the area coefficient of 18 means the model adds 18 currency units per additional square meter.

What do Ridge and Lasso add to linear regression?

Ridge and Lasso penalize coefficient size, which can help with many features or overlapping information. Lasso can also shrink some coefficients to zero.

How does a decision-tree regressor form predictions?

A decision-tree regressor repeatedly divides examples according to feature values and predicts a value for each resulting group.

How do random forests and gradient-boosted trees differ?

Random forests average predictions from many trees. Gradient-boosted trees build trees in sequence to improve predictions.

What does a regression residual measure?

A residual is the actual value minus the prediction: $y-\hat{y}$. A positive residual means the model predicted too low; a negative one means it predicted too high.

Name conditions that can make regression predictions unreliable.

Predictions may be unreliable when inputs differ greatly from training examples, important features are missing, or the data's patterns change.

How should you assess whether a regression model generalizes?

A prediction far outside the training-data range is an extrapolation and may not be realistic. Use held-out data to assess how well the model generalizes.

How does regression differ from classification?

Regression predicts a quantity that can vary across values; classification predicts a category.

What does ordinary least squares minimize?

Ordinary least squares chooses linear-model coefficients to minimize the sum of squared differences between observed and predicted targets.