The background of the Guess Who? game. What is a decision tree?
Today, there is a lot of talk about data analysis and decision trees, but…
What is a decision tree really?
It is difficult to truly understand what a decision tree is, but few of us realise that we have been using decision trees since childhood in something as simple as a game: Guess Who?
This game, whose rules and operation we all know, uses decision trees as follows.
If we have to guess that the opponent’s card is Jake, we will ask our fellow player the following questions:
- Is it a woman?
- Does he have a moustache/beard?
- Is he dark-haired?
- Does he have blond hair?
From which we would obtain the following answers:
- No
- Yes
- No
- Yes
From the first question, we would learn that the person we are looking for is a man; therefore, we could eliminate all the women. In other words, with a single question we have eliminated 6 possibilities from the game board.
With the second question, we learn that the eyes of the character we are looking for are brown; therefore, with this question, we are left with only 10 possible options.
Thanks to the third question, we learn that the person we are looking for is not dark-haired; therefore, we are left with only 4 options.
From the fourth and final question, we see that the only person remaining as a possible option is Jake.
Let us therefore see what the decision tree we followed to obtain the desired result would look like.
How do we apply decision trees to logistics?
This technique is easily applicable to many sectors and businesses. For example, if we are a courier company and we have to send a package of a certain size to a specific location, we can decide which van that will pass closest has enough space to carry this package. Therefore, we can assign a specific van to our package, one that fits the characteristics and requirements of each package.
In summary, we can say that decision trees, although they remain largely unknown in this data-saturated world with active trends such as Big Data or Business Intelligence, are a very useful tool with broad applications and one that greatly facilitates assigning a solution (as we saw in the previous example, the van) to a specific problem (delivering a package with specific requirements and characteristics).