Problem Description:

There are five attributes in the dataset Weather Dataset. Usually, the attribute "play" is treated as the target attribute and the other four the descriptive attributes.

In this lab, let's discard the attribute "play", and use the attributes "outlook", "humidity" and "windy" to predict the attribute "temperature".

Your tasks:

Build a decision tree to predict the temperature based on the descriptive attributes of outlook, humidity and windy.

Because the target attribute, temperature, is a numerical and continuous typed attribute, we have to use variance instead of entropy to measure the information gain.

Because the descriptive attribute, humidity, is also a numerical annd continuous typed attribute, we need to discretize it before use this attribute in the decision tree building.

There are so few data items (14) and so few descriptive attributes (3), you can even build the decision tree manually.