# Unit4, Section4: Decisions, Decisions

Instructional Days: 3

## Enduring Understandings

Decision trees are used to classify observations into similar groupings based on known characteristics. Questions are asked, then the observations are sorted based on the responses to the questions. After a specified number of iterations, a final group membership is decided. One particular modeling tool we use for decision trees is known as CART (Classification and Regression Trees).

## Engagement

Students will be presented with the question about whether they would rather trust a doctor or a data scientist to diagnose them if they were having a chest pains. This will set the context for decision trees and how they are used to make predictions.

## Learning Objectives

Statistical/Mathematical:

S-IC 2: Decide if a specified model is consistent with results from a given data-generating process, e.g., using simulation.

Data Science:

Understand that classification and regression trees can be used to predict membership in groups.

Applied Computational Thinking using RStudio:

• Create classification and regression trees.

Real-World Connections:

Cardiologists may use a decision tree to diagnose whether people are or are not having a heart attack. Since the late 1870’s, this method has been found to correctly diagnose a heart attack in over 95% of cases compared to correct diagnoses based on individual doctors’ expertise, which ranged between 75 and 90%.

## Language Objectives

1. Students will engage in partner and whole group discussions to express their understanding of classification trees.

2. Students will explain orally and in writing how to determine the accuracy of their non-linear model.

3. Students will make connections between decision trees and linear models in writing.

## Data File or Data Collection Method

Dataset:

1. USMNT/NFL dataset

Data File:

1. Titanic: data(titanic)

Data Collection:

Students will collect data for their Team Participatory Sensing campaign.