Revisitando a regressão linear

Linear regression fits a straight line that best predicts a number from your features. It is the simplest place to start any prediction. 📈

Every prediction comes from one formula: each feature gets a weight, and they add up. That weighted sum is the line the model draws through your data.

Those weights are called coefficients. Each one says how much the target moves when its feature goes up by one unit, holding the rest steady.

The intercept is the predicted value when every feature is zero. It shifts the whole line up or down to sit where the data lives.

Fitting picks the coefficients that make predictions closest to the real values. The model is just tuning the line until the errors shrink.

scikit-learn makes it tiny: create the model, then call fit with your features and target. The math happens for you.

After fitting, the learned weights live in coef_ and the offset in intercept_. They tell the story the model learned.

Hand fresh inputs to predict and the model applies the line to return numbers. Same simple call you saw with every estimator.

A positive coefficient means the target rises with that feature; a negative one means it falls. The sign is your first clue to the relationship.

Linear regression only bends in straight ways, so it can miss curvy patterns. Knowing this limit tells you when to reach for richer models.

It trains fast and is easy to explain, so it makes a great baseline. Beat this score before trusting anything fancier.

Let's confirm what each part of the fitted line means.

Linear regression draws a weighted line: coefficients set the slopes, the intercept anchors it, and fit tunes them to cut error. A clean, fast baseline. 🎯