Titanic Dataset – Exploratory Data Analysis

1. Dataset Structure

Schema Summary

Structure

The dataset contains 714 unique passengers with 8 attributes describing ticket class, demographics, and survival outcome.

Column	Type	Description (inferred)
survived	BIGINT	Target variable: `1` = survived, `0` = died
pclass	BIGINT	Passenger class: 1st, 2nd, or 3rd
name	VARCHAR	Passenger’s full name (includes title)
sex	VARCHAR	Biological sex: `male`, `female`
age	DOUBLE	Age in years (including fractional values for infants)
fare	DOUBLE	Ticket fare paid (in British pounds)
sibsp	BIGINT	Number of siblings and spouses aboard
parch	BIGINT	Number of parents and children aboard

Each row represents a unique passenger–there are 714 distinct names.

Global Metrics

Overview

Survival rate

40.6%

59.4% died

Class mix

~50% in 3rd class

1st: 26.1%, 2nd: 24.2%, 3rd: 49.7%

Sex distribution

63.4% male

36.6% female

Missing values

0 columns

No NULLs detected in any field

This Titanic extract is fully populated: no imputation is required for basic modeling and visualization.

2. Demographic and Ticket Distributions

Passenger Class & Sex

Categorical

The passenger list is dominated by third-class and male passengers, with roughly half of all travelers in third class and about two-thirds of passengers being men.

Class distribution

pclass	Passengers	Share
1st	186	26.1%
2nd	173	24.2%
3rd	355	49.7%

Sex distribution

sex	Passengers	Share
female	261	36.6%
male	453	63.4%

In visual form, bar charts of class and sex counts would quickly show the dominance of third-class and male passengers.

Age & Fare

Continuous

Ages range from infants to elderly passengers, while fares span several orders of magnitude and are strongly right-skewed.

Age summary

Metric	Age (years)
Min	0.42
Median	28.0
Mean	29.7
Max	80.0
Std. dev.	14.5

Fare summary

Metric	Fare
Min	0.00
Median	15.74
Mean	34.69
Max	512.33
Std. dev.	52.92

The large gap between median and mean fare reflects a handful of very high-priced tickets. In plots, a log-scale on fare makes the distribution easier to interpret.

3. Survival Patterns

Overall Survival

Target

Out of 714 passengers in this dataset, fewer than half survived. Survival is moderately imbalanced towards the negative class.

survived	Passengers	Share
0 (died)	424	59.4%
1 (survived)	290	40.6%

From a modeling perspective, this is a binary classification problem with a majority negative class.

Survival by Sex

Key Driver

Sex is one of the strongest predictors of survival. Women were heavily prioritized during evacuation.

sex	survived	Passengers	% within sex
female	0 (died)	64	24.5%
female	1 (survived)	197	75.5%
male	0 (died)	360	79.5%
male	1 (survived)	93	20.5%

Roughly three-quarters of women survived, while about four out of five men died—capturing the historical “women and children first” policy in the data itself.

Survival by Class

Socioeconomic

Passenger class shows a clear socioeconomic gradient in survival: higher classes had substantially better outcomes.

pclass	survived	Passengers	% within class
1st	0 (died)	64	34.4%
1st	1 (survived)	122	65.6%
2nd	0 (died)	90	52.0%
2nd	1 (survived)	83	48.0%
3rd	0 (died)	270	76.1%
3rd	1 (survived)	85	23.9%

About two-thirds of first-class passengers survived, compared with just one-quarter of third-class passengers. Cabin location and access to lifeboats are likely key mediating factors.

Class × Sex Interaction

Combined Effect

The combination of class and sex produces the most pronounced contrasts in survival rates.

pclass	sex	survived	Passengers	% within group
1st	female	0 (died)	3	3.5%
1st	female	1 (survived)	82	96.5%
1st	male	0 (died)	61	60.4%
1st	male	1 (survived)	40	39.6%
2nd	female	0 (died)	6	8.1%
2nd	female	1 (survived)	68	91.9%
2nd	male	0 (died)	84	84.8%
2nd	male	1 (survived)	15	15.2%
3rd	female	0 (died)	55	53.9%
3rd	female	1 (survived)	47	46.1%
3rd	male	0 (died)	215	85.0%
3rd	male	1 (survived)	38	15.0%

First-class women had an almost universal survival rate (~97%), while third-class men faced the worst outcomes (85% died). This interaction between class and sex provides a powerful narrative for visual storytelling (for example, as a heatmap).

4. Continuous Variables and Survival

Age vs. Survival

Demographic

Age differences between survivors and non-survivors are more subtle than those for sex or class, but still informative.

survived	Mean age	Median age
0 (died)	30.6	28
1 (survived)	28.3	28

Survivors are slightly younger on average, though the median age is identical. Density plots or age-band analyses (e.g., “children vs adults vs elderly”) would show that children tend to survive at higher rates.

Fare vs. Survival

Wealth Proxy

Ticket fare is a continuous proxy for wealth and cabin quality. Survivors, on average, paid substantially higher fares.

survived	Mean fare	Median fare
0 (died)	23.0	11.89
1 (survived)	51.8	26.25

Both mean and median fares for survivors are more than twice those of non-survivors, reflecting their concentration in higher classes. Box plots of fare by survival, especially on a log scale, visually reinforce this pattern.

5. Takeaways & Next Steps

Summary

Insights

This exploratory analysis confirms a clear, structured pattern in Titanic survival:

Sex and class are the dominant drivers of survival. Women, especially in first and second class, had very high survival rates, while third-class men fared the worst.
Fare and class align strongly with survival. Higher fares, typical of first-class cabins, are associated with much better outcomes.
Age plays a secondary but meaningful role. Survivors are slightly younger on average, with children more likely to survive than older adults, though the effect is weaker than that of sex or class.
No missingness simplifies downstream modeling. All variables are fully populated, so imputation is not required for this particular extract.

For modeling, useful next steps would include feature engineering (e.g. extracting titles from name, constructing family-size features from sibsp and parch), and training baseline classifiers to quantify the predictive power of these variables.