What Is Linear Regression - Least Squares Method

Quick reference

• 00:05 Okay, it's time to change things up a little bit and
• 00:07 talk a little bit about machine learning.
• 00:09 So we're not going to spend a whole lot of time talking about machine learning,
• 00:12 this is sort of the end of the course.
• 00:14 I just want to kind of whet your appetite and
• 00:16 show you a little bit of what's out there so you can start to think about what
• 00:19 you want to start learning next when it comes to data analysis with Python.
• 00:23 So we're going to be looking at linear regression,
• 00:26 which is sort of the most basic sort of intro thing in machine learning,
• 00:30 it's one of the easier models to do.
• 00:32 And we're going to walk through a linear regression in the next few videos,
• 00:35 like I said, we're not going to get into great detail.
• 00:37 In this video,
• 00:37 I just want to talk a little bit about what is linear regression.
• 00:40 So linear regression is a machine learning model prediction thing, right,
• 00:44 it allows us to make predictions of the future.
• 00:47 So think of it like if we know the height of different men, for instance, and
• 00:51 we know the average height of men, and a man has a child,
• 00:54 we can sort of use linear regression to predict how tall that child will be.
• 00:58 Things like that, just future things that we can predict numerically.
• 01:01 So if we head over to Wikipedia really quickly,
• 01:04 you can look up linear regression if you want to read about this.
• 01:07 Basically what we're doing is we're creating a scatterplot like this with
• 01:12 data, and we're passing a best fit line through it.
• 01:15 And then we can use that line to make predictions into the future.
• 01:19 So think of these as men's height for instance, right?
• 01:22 And given this data, if somebody up here has a kid, well,
• 01:26 we can sort of fit this line into the future to guesstimate how tall that
• 01:31 kid's going to be things like that, right?
• 01:34 So you have these points in this data, and
• 01:36 the thing you're trying to do is find the distance between the point and the line.
• 01:41 We call this distance the error, so we square this error,
• 01:44 and then we want to minimize that error in order to make predictions in the future.
• 01:48 So like I said, we're not going to get into this in great detail,
• 01:51 but that's basically what we're doing.
• 01:53 So we've done things like scatter plots before, we can we can do things like that.
• 01:57 So you may have heard of least squared errors method,
• 02:00 that's the same thing as linear regression.
• 02:03 And like I said,
• 02:03 we're just fitting a straight line just set of data points to make predictions.
• 02:06 So we need several tools in order to do this, we're going to use something called
• 02:11 Scikit-learn, which is a machine learning program for data analysis with Python.
• 02:15 We're going to be using something called Seaborn,
• 02:18 which is another charting system that's sort of like the Pandas and
• 02:21 Matplotlib we looked at in the last section.
• 02:23 Only the the charts and graphs are a little bit more fancy, but
• 02:26 they work in very similar ways to what we've already learned.
• 02:30 And we're also going to use SciPy to do some calculations and things like that.
• 02:33 So what we're going to be doing is running a linear regression on some
• 02:36 housing data from Boston.
• 02:38 So Scikit-learn comes with some sort of
• 02:41 dummy data that we can play around with and run our analysis.
• 02:45 So I'll walk you through just a basic linear regression analysis,
• 02:48 we'll take a look at it.
• 02:50 Hopefully, like I said, it'll whet your appetite, and show you sort of what's out
• 02:53 there, and what you could start to think about learning next after this course.
• 02:57 And like I said, we could spend hours and hours on machine learning,
• 03:00 we could spend hours and hours on linear regression, we're not.
• 03:03 We're going to spend 20 minutes or so, but it should be a good introduction and
• 03:07 kind of show you what's out there.
• 03:09 So in the next video, we will install all the things that we need to get started

Lesson notes are only available for subscribers.