Description
In this project we work with some toy data sets to compare the decision boundaries defined by
some of the simpler classifiers.
The project may be completed individually, or in a group of no more than three (3) people. All
students on the team must be enrolled in the same section of the course.
Platforms
The platform requirements for this project are the same as for previous projects.
Libraries
You will need pandas to load the datasets, scikit-learn to run the various classifiers, and
Matplotlib’s pyplot framework to visualize your results.
You may reuse code from the Jupyter notebooks accompanying the textbook and from the
documentation for the libraries. All other code and the results of experiments should be your
own.
Datasets
Download dataset1.csv, dataset2.csv, and dataset3.csv. These datasets have three
columns: features x1 and x2 and label t ∈ {0, 1}.
Experiments
Run the following experiments in a Jupyter notebook, performing each action in a code cell and
answering each question in a Markdown cell.
1. Use read_csv() to load and examine each dataset.
2. Use logistic regression to fit() and score() a binary classifier for dataset 1. How
accurate are the model’s predictions?
3. Repeat experiment (2) for dataset 2. How well does it score?
4. Create scatterplots for datasets 1 and 2, plotting points from class 0 with a different color
and marker from points in class 1. What accounts for the discrepancies between
experiments (2) and (3)?
5. Fit and score Gaussian Naive Bayes classifiers for datasets 1 and 2. How well do these
classifiers score compared to logistic regression?
6. Repeat experiment (5) with K-Nearest Neighbor classifiers.
7. Using the second half of the Python code for Figure 9.2 – Simple Gaussian Naive Bayes
Classification from Statistics, Data Mining, and Machine Learning in Astronomy, 2nd
Edition as a guide, plot the decision boundaries for each classifier and dataset. What
differences do you observe?
8. Now repeat experiments (2), (5), (6), and (7) with dataset 3.
Submission
Submit your Jupyter .ipynb notebook file through Canvas before class on the due date. Your
notebook should include the usual identifying information found in a README.TXT file.
If the assignment is completed by a team, only one submission is required. Be certain to identify
the names of all students on your team at the top of the notebook.