Linear Regression Quiz

Which of the following is least appropriate for linear regression?

• Predicting house prices

• Estimating someone's age from their Spotify playlist

• Modelling the relationship between study time and exam scores

• Generating random numbers

Answer explanation

Generating random numbers is least appropriate for linear regression, as this technique is used for modeling relationships between variables, not for creating random data without a defined relationship.

True or false: an R squared value of 0.85 means the model explains 85% of the variance in the dependent variable (the thing we are trying to predict)

• True

• False

Answer explanation

True. An R squared value of 0.85 indicates that the model explains 85% of the variance in the dependent variable, confirming the statement is correct.

What situation suggests your model is overfitting?

• It performs well on training data but poorly on new data

• The coefficients are all very close to zero

• R squared value is between 0.6 and 0.7

• The residuals are randomly scattered

Answer explanation

The correct choice indicates overfitting when a model excels on training data but fails to generalize to new data, highlighting its inability to capture the underlying patterns effectively.

What does the RMSE (Root Mean Squared Error) show?

• How many rows there are in the dataset

• The average size of the prediction errors, in the same units as the target

• The average size of the prediction errors, in the different units to the target

• The total number of prediction errors

Answer explanation

RMSE measures the average size of prediction errors, providing a clear indication of accuracy in the same units as the target variable. This makes it easier to interpret the model's performance.

What is a residual in linear regression?

• The amount of variation in the features

• The value that has been predicted

• The difference between the actual value and predicted value

• A type of feature scaling

Answer explanation

In linear regression, a residual is defined as the difference between the actual value and the predicted value. It measures how far off the predictions are from the true outcomes, making it a key concept in assessing model accuracy.

What does it mean if the residuals are spread out randomly around zero?

• The model is doing a good job

• The model is missing a pattern in the data

• The features have a lot of correlation

• There is an error somewhere

Answer explanation

If the residuals are spread out randomly around zero, it indicates that the model is capturing the underlying data well without systematic errors, meaning the model is doing a good job.

Draw a scatterplot that has no correlation.

What is the purpose of the fit() method in scikit-learn?

• To check for errors in the data

• To build the regression line based on training data

• To test the accuracy of the model

• There is no fit() method

Answer explanation

The fit() method in scikit-learn is used to build the regression line or model based on the training data provided. It adjusts the model parameters to best capture the underlying patterns in the data.

What does predict() do in a linear regression model?

• Trains the model on new data

• Creates a new feature

• Uses the model to estimate target values for new inputs

• Finds out which features are the most important

Answer explanation

The predict() function in a linear regression model uses the trained model to estimate target values for new input data, making it essential for making predictions based on the learned relationships.

Which of these is NOT an assumption of the linear regression model?

• The relationship between the features and the target is linear

• The residuals are normally distributed

• The features are strongly correlated with each other

• The variance of the residuals is constant (homoscedasticity)

Answer explanation

The correct choice is 'The features are strongly correlated with each other' because linear regression assumes that features are independent. The other options are valid assumptions of the linear regression model.