Titanic Passengers and Crew: Demographics and Survival Data

This dataset provides two tab-separated files for exploring Titanic survival by age and passenger class or crew membership: individual records for 2,207 people and an aggregated summary using three-year age bands. It is intended for teaching, descriptive analysis, and data visualisation. Files and format titanic.tsv: 2,207 person-level records, with five columns. titanic_aggregated.tsv: 100 age-band/class combinations, with five columns. Both files use UTF-8 encoding, tab separators, one header row, and LF line endings. Decimal values use a full stop. The missing-value marker NA is distinct from zero. Row counts exclude the header. Person-level data: titanic.tsv class: passenger class or crew membership; values are 1st, 2nd, 3rd, and Crew. sex: sex as recorded in the source; female or male. age: reported age in years at the sinking, including fractional infant ages. Two records have an unknown age, represented by NA. survived: survival outcome; 1 means survived and 0 means did not survive. embarked: embarkation location; B = Belfast, C = Cherbourg, Q = Queenstown, and S = Southampton. The file contains 1,317 passengers and 890 crew members: 324 first-class, 284 second-class, and 709 third-class passengers. There are 711 survivors and 1,496 people who did not survive. Recorded sex counts are 489 female and 1,718 male. The four source crew categories are combined into Crew, and the source field gender is named sex in this export. Names and identifiers are omitted. Identical rows are retained because different people can share the same five attributes. Ages have not been imputed or grouped in this file; the other columns contain no missing values. Rows are sorted by class (1st, 2nd, 3rd, Crew), sex (female, male), numeric age ascending, survival outcome (0, 1), and embarkation code alphabetically. Missing ages follow known ages within each class/sex group. Aggregated data: titanic_aggregated.tsv age_midpoint: integer age-band label, taking the values 1, 4, 7, …, 73. class_code: 1 = first class, 2 = second class, 3 = third class, and 4 = crew. survival_pct: 100 × survivors / n, rounded to six decimal places; NA when the group is empty. n: number of people with known age in the group. survivors: number of survivors in the group. The summary crosses 25 age bands with four class categories. Bands are [0,3), [3,6), …, [72,75): each includes its lower bound and excludes its upper bound. Both sexes are pooled. The two people with unknown ages are excluded, leaving 2,205 people and 711 survivors. Despite its name, age_midpoint contains the middle completed-year age, rather than the exact midpoint of the continuous interval. For example, label 1 represents ages 0–2, including fractional ages below three; label 4 represents ages 3–5. No individual ages are rounded before grouping. Empty combinations are retained with n = 0, survivors = 0, and survival_pct = NA. They must not be interpreted as zero-percent survival. Rows are ordered by increasing age-band label and then class code. All 100 combinations, group counts, survivor counts, and percentages have been checked against the person-level file. Source and preparation The preparation notes identify the direct source as the Titanic CSV in Dennis Sun’s Principles of Data Science repository, accessed on 25 September 2026. Selected fields were manually transcribed from the browser-readable CSV and exported programmatically; the person-level export was subsequently sorted as described above. The source URL refers to a mutable branch rather than a pinned revision. A related 2,207-person dataset is documented in the stablelearner reference manual. The exact transformation history linking that dataset to the educational CSV has not been independently established. This collection contains individual records and is distinct from the familiar 2,201-person Titanic contingency table. Interpretation and limitations These are descriptive data for the supplied cohort, not a definitive historical reconstruction. Reported ages may be uncertain. A complete record-by-record comparison with the upstream CSV has not been performed, so transcription errors remain possible despite internal consistency checks. Some age/class groups contain very few people; percentages should therefore be interpreted alongside n. The summary contains no uncertainty estimates. Class codes identify categories, and crew is not a fourth passenger class. Pooling sexes and crew roles can conceal differences within groups. The data alone do not establish causal effects of age or class on survival.