📃 Solution for Exercise M1.02

The goal of this exercise is to fit a similar model as in the previous notebook to get familiar with manipulating scikit-learn objects and in particular the .fit/.predict/.score API.

Let’s load the adult census dataset with only numerical variables

import pandas as pd

adult_census = pd.read_csv("../datasets/adult-census-numeric.csv")
data = adult_census.drop(columns="class")
target = adult_census["class"]

In the previous notebook we used model = KNeighborsClassifier(). All scikit-learn models can be created without arguments. This is convenient because it means that you don’t need to understand the full details of a model before starting to use it.

One of the KNeighborsClassifier parameters is n_neighbors. It controls the number of neighbors we are going to use to make a prediction for a new data point.

What is the default value of the n_neighbors parameter?

Hint: Look at the documentation on the scikit-learn website or directly access the description inside your notebook by running the following cell. This opens a pager pointing to the documentation.

from sklearn.neighbors import KNeighborsClassifier

KNeighborsClassifier?

We can see that the default value for n_neighbors is 5.

Create a KNeighborsClassifier model with n_neighbors=50

# solution
model = KNeighborsClassifier(n_neighbors=50)

Fit this model on the data and target loaded above

# solution
model.fit(data, target)

KNeighborsClassifier(n_neighbors=50)

In a Jupyter environment, please rerun this cell to show the HTML representation or trust the notebook.
On GitHub, the HTML representation is unable to render, please try loading this page with nbviewer.org.

Use your model to make predictions on the first 10 data points inside the data. Do they match the actual target values?

# solution
first_data_values = data.iloc[:10]
first_predictions = model.predict(first_data_values)
first_predictions

array([' <=50K', ' <=50K', ' <=50K', ' <=50K', ' <=50K', ' <=50K',
       ' <=50K', ' >50K', ' <=50K', ' <=50K'], dtype=object)

first_target_values = target.iloc[:10]
first_target_values

0     <=50K
1     <=50K
2     <=50K
3     <=50K
4     <=50K
5     <=50K
6     <=50K
7      >50K
8     <=50K
9      >50K
Name: class, dtype: object

number_of_correct_predictions = (
    first_predictions == first_target_values
).sum()
number_of_predictions = len(first_predictions)
print(
    f"{number_of_correct_predictions}/{number_of_predictions} "
    "of predictions are correct"
)

9/10 of predictions are correct

Compute the accuracy on the training data.

# solution
model.score(data, target)

0.8290379545978042

Now load the test data from "../datasets/adult-census-numeric-test.csv" and compute the accuracy on the test data.

# solution
adult_census_test = pd.read_csv("../datasets/adult-census-numeric-test.csv")

data_test = adult_census_test.drop(columns="class")
target_test = adult_census_test["class"]

model.score(data_test, target_test)

0.8177909714402702

Looking at the previous notebook, the accuracy seems slightly higher with n_neighbors=50 than with n_neighbors=5 (the default value).