案例学习：预测房价

Loading...

來自 University of Washington 的課程

机器学习：回归

3717 個評分

案例学习：预测房价

從本節課中

Multiple Regression

The next step in moving beyond simple linear regression is to consider "multiple regression" where multiple features of the data are used to form predictions. <p> More specifically, in this module, you will learn how to build models of more complex relationship between a single variable (e.g., 'square feet') and the observed response (like 'house sales price'). This includes things like fitting a polynomial to your data, or capturing seasonal changes in the response value. You will also learn how to incorporate multiple input variables (e.g., 'square feet', '# bedrooms', '# bathrooms'). You will then be able to describe how all of these models can still be cast within the linear regression framework, but now using multiple "features". Within this multiple regression framework, you will fit models to data, interpret estimated coefficients, and form predictions. <p>Here, you will also implement a gradient descent algorithm for fitting a multiple regression model.

- Emily FoxAmazon Professor of Machine Learning

Statistics - Carlos GuestrinAmazon Professor of Machine Learning

Computer Science and Engineering

[MUSIC]

So the first approach that we're gonna talk about is just a closed form solution

where we take our gradient, and simply set it equal to zero, and solve for w.

Just like we did in the simple regression case.

Okay, here's my gradient of our residual sum of squares, and

I've set it equal to zero.

And now let's solve for w.

So, let's do out this multiplication here where we're gonna get -2H transpose, y.

And then we're gonna get a +2H transpose,

H times w, and we're setting this equal to zero.

And so these matrix multiplies act very similarly to if these had been scalars,

but we have to keep the order.

The order is very important.

We couldn't have switched it around to be y h transposed, okay.

So, now let's solve this equation.

First of all, the two cancels out, we can just divide both sides by two.

And so what I'm gonna end up with is h transpose

hw equals, I bring this to the other side, h transpose y.

Then I'm gonna multiply both sides by h transpose h inverse.

So H transpose Hw and now I'm gonna multiple,

pre-multiply the other said by the same.

H transpose H inverse, H transpose y and so this is a little aside.

What is a matrix A inverse times the same matrix a.

Well, that's the definition of,

so, by definition of a matrix inverse that is the identity matrix.

And another aside, so that's aside number one another aside if I take

a the identity matrix and multiply by any vector,

so I'll call the vector v, I just get v back.

Or if I take the matrix and multiply it by any matrix, big V,

I'm gonna get that matrix back.

Okay, so what we see if we apply these two identities here is that together,

these two terms, h transpose h is like our big A matrix here.

So we have a matrix times it's inverse, this is just gonna be the identity.

Then we have the identity matrix times a vector w,

that's just gonna be the w vector.

So we have w, and let me put hats on, because remember,

that once we set this equal to zero and solve for w this is our

estimated set of coefficients so I'm going to put a hat on w and

what we see is w hat is simply equal to h transpose h inverse HTy.

So this is one of those aha moments.

Maybe you missed it.

Cuz aha moment is, we have a whole collection of different parameters,

w0 all the way up to wd.

These are the things multiplying all the features we're using in our multiple

regression model.

And in one line I've written the solution to the fit.

I've fit all of w 0, w 1, all the way up to D just by doing this matrix multiply.

Okay?

So this.

Motivates why we went through all this work to write things in this matrix

notation because it allows me to have this nice closed form solution for

all my parameters written very compactly.

[MUSIC]