Consider the following training data set for $10$ instances. Each instance has four features: color, size, texture and number, and is categorized as either a positive or a negative $($the type column, which is the target value$).$

Using the decision tree approach we saw in class, the feature chosen first to split the data set

on is:

texture

size

color

number

There is a tie between two features, $,$ either of which can be chosen.