Data Types
Data types recognized by DuckDB and how they appear in ExploreMyData.
| Type | Examples | Explorer Card |
|---|---|---|
VARCHAR (text) | "hello", "2024-01-01", "true" | Value counts with bar chart |
INTEGER / BIGINT | 42, -7, 1000000 | Histogram |
DOUBLE (decimal) | 3.14, -0.5, 99.99 | Histogram |
BOOLEAN | true, false | Value counts. Booleans are grouped with text columns, so they get the categorical card, equality filters, and search rather than histograms or SUM/AVG. |
DATE | 2024-01-15 | Time-series chart |
TIMESTAMP | 2024-01-15 10:30:00 | Time-series chart |
TIME | 10:30:00 | Value counts. Displayed as HH:MM:SS. |
STRUCT / LIST | {"a": 1}, [1, 2, 3] | Value counts. Displayed as JSON. |
Type Detection
DuckDB automatically detects column types when loading your file. It samples the data and chooses the most specific type that fits all values.
Type Conversion
Use the Convert Type operation to change column types. DuckDB uses TRY_CAST which returns null instead of erroring on invalid values.
Converting text to a number handles the formatting real files carry. Thousands separators, currency symbols and percent signs are stripped first, and values wrapped in parentheses are read as negative, so 1,234.56, $500, (500) and 5% all convert cleanly. The converted column keeps its original position in the grid.