What kind of data can I visualize with RAWGraphs 2.0?

Updated

RAWGraphs 2.0 works best with tabular data — rows of records and columns of fields, typically pasted from a spreadsheet or loaded from a CSV or delimited text file. You then map those columns onto visual variables (x, y, color, size, shape) to build exploratory charts. It is a visualization tool, not a statistics package, so it suits tasks where you want to see structure in data rather than run tests or models.

The data shape it expects

RAWGraphs assumes a flat table: one row per item, one column per attribute. That is the same layout you get from Excel, Google Sheets, or a database export.

Your data Works well? Why
CSV / TSV / semicolon-delimited text Yes Native tabular input
Spreadsheet copy-paste Yes Rows and columns map directly
One row per record, one column per field Yes Matches the mapping model
Nested JSON or XML Not directly Needs flattening first
Multi-sheet workbooks with merged cells Not directly Needs a clean single table
Time series with one column per date Awkward Better reshaped to long format

If your file has a header row, keep it — column names become the labels you pick from when mapping. Keep one value per cell; merged cells and multi-line cells tend to break the import.

What you can map

The core idea is that each column becomes a visual channel. Typical mappings:

  • Category column → x or y position (e.g., country, department)
  • Numeric column → size or length (e.g., population, revenue)
  • Another category → color (e.g., region, segment)
  • A third dimension → shape, or a small-multiple panel

This is why RAWGraphs is described as exploratory: you try a mapping, look at the result, and change it. The tool is built around that loop rather than around a fixed chart type.

Tasks it fits

  • Comparing categories — bar-like and circular layouts where a category drives position and a number drives length.
  • Showing composition — stacked or nested layouts where parts sit inside a whole.
  • Revealing relationships — scatter-style layouts with two numeric axes plus color or size for a third field.
  • Hierarchies and flows — tree and flow layouts, provided the data is already in the parent/child or source/target form the layout needs.
  • Quick exploratory passes — pasting a table, trying several mappings, and keeping the one that reads clearly.

Tasks it does not fit

  • Statistical analysis — no regression, significance testing, or model fitting.
  • Live or streaming data — you load a snapshot, not a connection.
  • Heavy data cleaning — reshape and fix your table before importing.
  • Large datasets — browser-based rendering has practical limits; very wide or very long tables get slow.

Before you import: a quick checklist

  1. One header row, one field name per column, no duplicate names.
  2. One value per cell — no merged cells, no line breaks inside cells.
  3. Consistent delimiters; if commas appear inside values, quote those values or use a different delimiter.
  4. Numeric columns stored as numbers, not text with units attached.
  5. Long format rather than wide when you have repeated measures across dates.

Output and what happens next

Charts are produced in the browser and can be exported for presentation or further editing in a vector or design tool. Treat the export as a starting point for a final graphic rather than a finished publication asset — labels, legends, and annotation usually need a pass afterward.

One practical note: the app requires JavaScript to run, so if you see only a blank page, check that scripts are enabled and that nothing is blocking them.