Decision trees
This class of algorithms aims to predict the unknown labels splitting the dataset, by generating a set of simple rules that are learnt from the features values. For example, consider a case of deciding whether to take an umbrella today or not based on the values of humidity, wind, temperature, and pressure. This is a classification problem, and an example of the decision tree can be like what is shown in the following figure based on data of 100 days. Here is a sample table:
|
Humidity (%) |
Pressure (mbar) |
Wind (Km/h) |
Temperature (C) |
Umbrella |
|---|---|---|---|---|
|
56 |
1,021 |
5 |
21 |
Yes |
|
65 |
1,018 |
3 |
18 |
No |
|
80 |
1,020 |
10 |
17 |
No |
|
81 |
1,015 |
11 |
20 |
Yes |
Decision tree for predicting whether to bring an umbrella or not based on a record of 100 days.
In the preceding figure the numbers in squares represent the days on which an umbrella has been brought, while the circled numbers indicate days in which an umbrella was not necessary.
The decision tree presents two types of nodes: decision nodes, which...
