Genealogy Data Analysis: Statistics, Maps and Patterns from Your GEDCOM File

Genealogy software treats your file as a tree to be navigated. Data analysis treats it as a dataset to be queried. The distinction matters because the most interesting questions about a family are statistical — patterns over time, geographic concentrations, demographic trends — and a tree view can't answer them.

GEDminer parses your GEDCOM into structured tables in memory: one row per individual, one per family, one per fact, one per source, one per place. Every analytical view is then a pivot over those tables. Birth-year histograms count rows per decade. Surname distributions group rows by family name. Migration maps plot edges between birth and later residence locations. Occupation breakdowns categorise the OCCU tag values into historical groups.

Because the analytics are computed from the dataset, every view stays consistent: the count of individuals on the dashboard equals the row count in the People tab equals the total in the surname histogram. There is no manual aggregation to drift out of date.

Anyone running a one-name study needs three things: an accurate frequency count for the surname (and its variants) across time, a geographic distribution, and a way to spot related branches that may not yet be merged.

GEDminer's surname tools provide all three:

• Frequency by decade shows how the surname's prevalence in your tree has changed over time.
• Variant clustering uses Soundex + Levenshtein with a 0.55 similarity threshold to group spellings (Smith / Smyth / Smithe) so the count reflects the true population, not the spelling fragmentation.
• Geographic concentration plots the surname onto a centroid map per decade, so you can watch a family migrate.

Combined with the Migration Analysis and the Hidden Cousin Connector, this is enough to drive most of the analytical work in a serious one-name study.

Migration is computed at two levels: per-individual and aggregated.

Per individual, the analyzer compares an individual's birth, marriage, residence and death locations. Each pair of consecutive different locations becomes a migration "leg" with a year (or year range) attached. Legs that cross a national border become flagged as international moves.

Aggregated, all legs are bucketed by decade and by origin/destination country, producing flow lines on the map and the per-decade peak-migration histogram. Internal moves (within a country) are shown separately from international ones so you can answer questions at either level.

The map deliberately strips most labels for visual clarity; you can drill into any country or region to see the individual ancestors driving the flow.

Raw GEDCOM occupation tags are messy: "ag lab", "agricultural labourer", "Ag. Lab.", "farm worker" and "labourer (farm)" are all the same thing recorded five different ways.

GEDminer normalises occupation strings (case folding, abbreviation expansion, trailing-punctuation stripping) and assigns each one to a historical category: Agriculture, Trades & Crafts, Industrial, Professional & Clerical, Domestic Service, Military, Maritime, Religious, Mining & Extractive, Transport, Retail & Commerce, and Unclassified.

Categories are mapped consistently per era so you can compare an 18th-century "weaver" with a 19th-century "factory weaver" without losing the historical distinction. The breakdown is shown both as a flat distribution and as a stacked-bar over time.

Once you have birth and death dates for a critical mass of ancestors, demographic patterns fall out of the data. GEDminer surfaces several:

• Mean and median lifespan by birth decade — typically rising from the late 19th century onwards.
• Family size by marriage decade — typically falling after about 1870.
• Age at first marriage — historically lower than people often assume.
• Sibling spacing — useful for catching incomplete families.

These are computed from the underlying data tables, so they update automatically as you add records. They're as much a sanity check as a research tool — anomalies (lifespans suddenly halving in a decade, family sizes spiking unrealistically) usually point to an underlying data quality issue worth investigating.