Search⌘ K
Home
Courses
Hands-On Quantum Machine Learning with Python

The Convenience Function and Probabilities for Relatives

Explore how passenger relationships influence survival outcomes by analyzing familial ties in the Titanic dataset. Learn to use convenience functions to evaluate probabilities based on relatives aboard and improve quantum classification models. This lesson introduces methods to identify likely family groups and apply these insights to refine predictions.

We'll cover the following...

How familial relationships affect survival

Let’s get back to the Titanic. There are still plenty of ways to improve our QBN. A promising feature to include is the relationships between passengers. So far, we’ve ignored any family relationship. Our dataset contains information about the number of siblings and spouses (SibSp) and the number of parents and children (Parch) traveling with a passenger.

The SibSp and Parch fields are numeric values denoting the number of related passengers aboard the Titanic.

The following function lets us evaluate how a certain number of related passengers affects the chance to survive.

Convenience function to evaluate the effect of a relationship

Javascript (babel-node)
def evaluate_relation(relation, value):
    # separate the population
    population = train[train[relation].eq(value)] if value < 2 else train[train[relation].ge(value)]
    p = len(population)/len(train)
    # chance to survive
    surv = population[population.Survived.eq(1)]
    p_surv = len(surv)/len(population)
    return (p, p_surv)

The function evaluate_relation takes two parameters, the name of the relation and the value. We’ll start by separating the population from our training dataset in line 3. If the provided value is smaller than 2, we select all passengers with this exact value for the given relation. We summarize all passengers with a value that is greater or equal to 2. The marginal probability of having a certain number of related passengers is given by the size of the selected population divided by the size of the training dataset in ...