Linear Regression F23

Supervised learning maps or relates...

• input data --> labeled output data

• input data --> unlabeled output data

• output data --> labeled input data

• output data --> unlabeled input data

Answer explanation

This is the definition! In supervised learning, we give our model lots of examples of labeled inputs --> labeled outputs. Our final goal is to be able to map an unlabeled input to a labeled output!

In unsupervised learning, our model is given inputs, and the model then has to figure out how to group those inputs on its own.

How can we avoid overfitting?

• Use a LESS diverse dataset

• Use more data

• Use a MORE diverse dataset

• Use less data

Answer explanation

Overfitting happens when our model keeps seeing the same pattern again and again -- and memorizes the pattern instead of generalizing it! More data means that we have more examples to generalize off of, and more diverse data means that our model sees different patterns and won't only memorize the same thing.

How can we avoid underfitting?

• Use a model that guesses more randomly

• Use more data

• Use less data

• Make our model more powerful/expressive

Answer explanation

Underfitting is caused by our model not having enough information, or when our model is not complex enough to capture relationships between the inputs and outputs.

Regression refers to the relationship between two or more variables.

• True

• False

Answer explanation

Yes! this is the definition of regression

Is this an example of linear regression?

• Yes

• No

Answer explanation

Yes, this is linear regression! We get a straight line that is approximating the relationship between inputs and outputs of a scatterplot

Is this an example of linear regression?

• Yes

• No

Answer explanation

Is this an example of linear regression?

• Yes

• No

Answer explanation

This is a "straight line"! This is linear regression in multiple dimensions. Imagine the axis going up is our housing price, the axis going to the right is the number of rooms, and the axis going into the page is square feet.

• x: house location, y: house size

• x: house size, y: house location

• x: house location, y: house price

• x: house size, y: house price

Answer explanation

Remember, y represents the OUTPUT, which is house price!

x represents the INPUTS, which could be a VECTOR of house size, house location, and other factors