Tabular data is the most common format encountered in business and scientific computing.
It structures observations into a rectangular grid where geometry carries precise meaning.
Horizontal rows are often called records, tuples, instances, or observations.
Vertical columns are often called attributes, features, fields, or variables.
The intersection of a single row and a single column holds one specific data value.
Each column must maintain a consistent meaning and data type down its entire length.
If an age column contains numbers, placing a person's city in that column corrupts the structure.
Rows should be uniquely identifiable, often through a designated identifier column or primary key.
This grid design allows computers to search, filter, and calculate values across millions of entries.
Confusion arises when people use rows for attributes and columns for individual dates or entities.
That rotated layout makes mathematical summaries and database queries cumbersome and slow.
Consistent rectangular structure is required before running mathematical operations or training algorithms.
In spreadsheets, people often break tables by embedding summary rows or blank spacer rows.
Clean tabular design strictly separates raw data tables from presentation views and summary reports.
Respecting the row-entity and column-attribute rule prevents countless data processing headaches.