Scribd
Upload
30 views3 pages

Lab 2

Uploaded by

ptyquyen22
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
Download as pdf or txt
30 views3 pages

Lab 2

Uploaded by

ptyquyen22
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
Download as pdf or txt
Download as pdf or txt
You are on page 1/ 3

Artificial Intelligence

Lab 2: Decision Tree


1 Description

1.1 Description
In this assignment, you are going to build a decision tree on the UCI Nursery Data Set, with support from the
scikit-learn library
There are 12960 records in the data set. Nursery Database was derived from a hierarchical decision model originally
developed to rank applications for nursery schools.
You can download the dataset here: https://archive.ics.uci.edu/dataset/76/nursery

1.2 Assignment requirements


You are asked to write a Python program and use scikit-learn functions to fulfill the following tasks.
Although there is no strict rule on organizing the code, each task should be noted carefully and must reflect all the
requirements mentioned.

1.2.1 Preparing the data sets

This task prepares the training sets and test sets for the incoming experiments. Add the ”.csv” extension to the file
”nursery.data” to read the file
You prepare the following four subsets from the data in ”nursery.data.csv”:

• feature_train: a set of training examples, each of which is a tuple of 8 attribute values (target attribute
excluded).

• label_train: a set of labels corresponding to the examples in feature_train.

• feature_test: a set of test examples, it is of similar structure to feature_train

• label_test: a set of labels corresponding to the examples in feature_test.

You need to shuffle the data before splitting and split it in a stratified fashion. Other parameters (if there are any)
are left by default.
There will be experiments on training sets and test sets of different proportions, including (train/test) 40/60, 60/40,
80/20, and 90/10; thus, you need 16 subsets.
Visualize the distributions of classes in all the data sets (the original set, training set, and test set) of all proportions
to show that you have prepared them appropriately

1.2.2 Building the decision tree classifiers

This task conducts experiments on the designated train/test proportions listed above.
You need to fit an instance of sklearn.tree.DecisionTreeClassifier (with information gain) to each training set and
visualize the resulting decision tree using graphviz.
Artificial Intelligence

1.2.3 Evaluating the decision tree classifiers

For each of the above decision tree classifiers, predict the examples in the corresponding test set, and make a report
using classification_report and confusion_matrix.

How do you interpret the classification report and the confusion matrix? From that, make your comments on the
performances of those decision tree classifiers.

1.2.4 The depth and accuracy of a decision tree

This task works on the 80/20 training set and test set. You need to consider how the decision tree’s depth affects the
classification accuracy.
You can specify the maximum depth of a decision tree by varying the parameter max_depth of the decision tree.
You need to try the following values for parameter max_depth: None, 2, 3, 4, 5, 6, 7. And then,

• Provide the decision tree drawn by graphviz for each max_depth value

• Report to the following table the accuracy_score (on the test set) of the decision tree classifier when changing
the value of parameter max_depth.

max_depth None 2 3 4 5 6 7
Accuracy

• Make your comment on the above statistics.


Artificial Intelligence

2 Grading

No. Specifications Scores (%)


1 Preparing the data sets 30
2 Building the decision tree classifiers 20
3 Evaluating the decision tree classifiers
Classification report and confusion matrix 10
Comments 10
4 The depth and accuracy of a decision tree
Trees, tables, and charts 20
Comments 10

3 Notice
• This is an INDIVIDUAL assignment.

• You must use Python language and present the code in a single ipynb file.

• Write down your report on a PDF File.

• All the required visualizations must be presented in the ipynb file, while statistical results and comments are
presented in the report.

• A program with syntax/runtime error(s) will not be accepted.

You might also like