Name: Writing Our First Classifier - Machine Learning Recipes #5
Uploaded: 2017-07-17T11:49:47.000Z
Duration: 8 min 44 s
Description: Thousands of YouTube videos with English-Chinese subtitles! Now you can learn to understand native speakers, expand your vocabulary, and improve your pronunciation...

In this episode, we're going to do something special,

and that's write our own classifier from scratch.

Because if you can follow along and do this on your own,

it means you understand an important piece of the puzzle.

The classifier we're going to write today

is a scrappy version of k-Nearest Neighbors.

That's one of the simplest classifiers around.

First, here's a quick outline of what we'll do in this episode.

We'll start with our code from Episode 4, Let's

Recall in that episode we did a simple experiment.

We imported a data set and split it into train and test.

Here we'll comment them out and write our own.

The rest of the pipeline will stay exactly the same.

I'll pop in and out of the screencast to explain things

To start, let's run our pipeline to remind ourselves

So the first thing we need to do is fix our pipeline.

And to do that, we'll implement a class for our classifier.

Next, I'll change our pipeline to use it.

Now let's see what methods we need to implement.

Looking at the interface for a classifier,

we see there are two we care about-- fit,

Remember this takes the features and labels for the training set

as input, so we'll add parameters for those.

As input, this receives the features for our testing data.

And as output, it returns predictions for the labels.

Our first goal is to get the pipeline working,

And by random, I mean we'll just guess the label.

To start, we'll add some code to the fit and predict methods.

In fit, I'll store the training data in this class.

You can think of this as just memorizing it.

that we'll need to return a list of predictions.

That's because the parameter, X_test, is actually

Each row contains the features for one testing example.

I'll just randomly pick a label from the training data

At this point, our pipeline is working again.

So let's run it and see how well it does.

Recall there are three different types of flowers

in the iris dataset, so accuracy should be about 33%.

Now we know the interface for a classifier.

To do that, we'll implement our classifier,

Here's the intuition for how that algorithm works.

Let's return to our drawings of green dots and red dots

are the training data we memorized in the fit method,

Now imagine we're asked to make a prediction for this testing

For example, we'll guess that this testing dot is green,

because that's the color of its nearest neighbor.

As another example, if we had a testing dot over here,

Now what about this one right in the middle?

Subtitles ListPlay Video

Writing Our First Classifier - Machine Learning Recipes #5

episode

random

term

comment