The purpose of a decision tree is to visualise features of a model by means of a tree-like graph, and infer the importance (and lack thereof) of each feature in affecting the output variable.

The decision tree structure consists of:

• Nodes: Each decision tree consists of what are called root nodes and decision nodes.
• Branches: Represent the outcome of each decision taken across the nodes.

Source: Photo by OpenClipart-Vectors from Pixabay.

In this example, a regression-based decision tree is formulated to predict ADR (average daily rate) for a hotel given certain customer attributes.

Background

This study focuses on hotel booking analysis. When it comes to hotel bookings, average daily rate (ADR) is a particularly important metric. This reflects the average rate per day that a particular customer pays throughout their stay.

This analysis is based on the original study by Antonio, Almeida, and Nunes, 2016.

Gauging ADR allows hotels to more accurately identify its most profitable customers and tailor its marketing strategies accordingly.

The chosen features that form the input for this neural network are as follows:

1. IsCanceled
2. Country of origin
3. Market segment
4. Deposit type
5. Customer type
6. Required car parking spaces
7. Arrival Date: Year
8. Arrival Date: Month
9. Arrival Date: Week Number
10. Arrival Date: Day of Month

#regression #python #machine-learning #data-science #decision-tree