Did First Class Passengers really have a better Chance of Surviving the Titanic Disaster?
Exploratory Data Analysis with Python 101 : Survivability Study of the Titanic Disaster
Introduction
Exploratory Data Analysis (EDA) is a crucial process that allows us to dive deeper into a dataset and discover the stories it holds. By exploring patterns, relationships, and outliers, EDA helps data scientists to spot anomalies, test hypotheses, and check assumptions, providing a better understanding of the dataset (1).
In this article, I’ll guide you through an EDA of the Titanic dataset, exploring intriguing questions like: Did class or gender play any significant role in survival? How much did fare and age really impact who made it off the ship?
Tutorial objectives
The objective of this EDA tutorial is to share techniques you can apply to explore any dataset, specifically using Python. It follows a hands-on approach with Python libraries like Pandas, Seaborn, and Matplotlib.
We will approach this in an engaging way by verifying the following famous assumption about the Titanic disaster: it is believed that passengers who paid more for their tickets had a better chance of survival. This is primarily because first-class passengers, often located on higher decks, had better access to lifeboats and evacuation routes, which improved their survival chances. In the other hand, lower-class passengers, situated on lower decks, faced more challenges in reaching lifeboats, particularly as water filled the lower sections of the ship first, resulting in lower survival rates. However, there was an exception for children, as many young passengers were prioritized during evacuations, increasing their chances of survival regardless of their class (2).
Tutorial dataset
For this analysis, we'll be using the Titanic dataset available on Kaggle, which provides information on the fate of passengers aboard the Titanic, categorized by class, age, gender, and other attributes. I highly recommend to check the Data Dictionary, to understand better the meaning of each column of the dataset.
In order to verify our assumption, I will conduct the following explorations to gain a better understanding of the factors influencing survival rates:
Let’s dive into the analysis and uncover what the data has to say.
Results and Python source code:
The EDA is available on my GitHub as a Jupyter notebook, which outlines each step of the analysis and provides conclusions for each section. You can directly continue reading on the notebook, for the rest of the tutorial.
领英推荐
Tutorial breakdown and outcomes
This is the most important part of the tutorial, where you will be learning some Python skills for exploring data.
We will start by visualizing the distribution of key features and analyzing the relationship between fare and class, establishing fare as a strong indicator of class using correlation and the P-value.
Next, we will use the "groupby" method on different categories, this will help us explore data efficiently and identify some patterns. For instance, we will check the gender distribution by class, revealing that women were more prevalent in first class while men dominated third class. We will examine survival rates by class, demonstrating how socioeconomic status affected survival, and analyze survival rates by gender, showing that females had significantly better chances than males.
Finally, we will visualize the impact of age on survival, noting that children had the highest survival rates, while seniors had the lowest, regardless of their class.
Wrap up
The analysis shows that survival rates in the Titanic disaster involve many major factors that include: Socioeconomic status, as represented by class, has an enormous effect on the chances of survival; The data confirmed the argument that higher-class passengers had better access to lifeboats, which saved their lives. Second, gender was another critical determinant, as women, especially in first class, had higher survival rates. Thirdly, the age contributed in some exceptions, with children having relatively high survival rates while seniors had very poor chances, regardless of the class of these two age groups.
Overall, these findings show that class, gender, and age are interconnected in their impact on survival outcomes in the Titanic disaster, reflecting how social structures shaped the course of individual fates in a time of crisis.
By the end of this article, you have learned how you can perform EDA on any dataset using Python. You are now able to explore new datasets, draw meaningful insights through visualization, distribution techniques, variable relationships analysis, and missing data handling.