How to plot CSV data on a Mac

Quick path: CSV to graph in six steps

1. Open the CSV, TSV or other delimited text file in a tool that shows a preview of the parsed table.
2. Set the delimiter, header row and number of rows to skip until every field sits in its own column and the first real observation is still present.
3. Rename columns for the quantity and unit they hold, and decide what an empty cell, NA and zero each mean.
4. Plot the independent variable on X and the response on Y; add further series only if they share units and X definition.
5. Check range, gaps, outliers, scale and error values on the plot itself.
6. Save the project or script with the source file, then export at the final size.

The rest of this guide explains what to check at each step and why.

Check the CSV import preview first

CSV is a loose convention, not one format. Files differ in delimiter (comma, semicolon, tab, space), decimal separator, quoting, preamble lines from instruments, and whether the first line is a header. A file renamed to .csv keeps its original structure.

In the preview, confirm that:

• each expected field occupies exactly one column, and the column count matches what the instrument wrote;
• the header is a header, not the first observation, and no units row became data;
• decimal values are numeric, not text, and no thousands separators split a number;
• comment or metadata lines at the top are skipped, and the first real observation is still there;
• the row count matches the file, so nothing was silently dropped.

Keep one row per observation and one column per variable where you can. Units belong in the column name, not in numeric cells.

Name variables, then choose X and Y

A plot inherits its labels, legend and provenance from the column names. Rename before you plot: pressure_kPa saves a later edit in every figure and makes the exported data readable by someone else.

Put the controlled or independent quantity on X and the measured response on Y. Many tools, Autoplot included, fall back to row index when X is left empty. That is useful for a quick look at a sequence, but it assumes evenly spaced, correctly ordered samples, which instrument logs often are not.

For several related CSV files, the same rules apply per file before anything is combined. The guide to plotting multiple CSV files covers merging and comparing runs.

Inspect the graph before you interpret it

Start with the points, not a fitted line. Read the full range before zooming, and look for gaps, repeated X values, isolated spikes and values that are exactly zero where a measurement should be.

• Scale: use a logarithmic axis when the data span orders of magnitude or the model is multiplicative, not to make a curve look straighter. Non-positive values cannot appear on a log axis, so check what was dropped.
• Error bars: add them only when you know whether they show a standard deviation, a standard error or an instrument tolerance, and say which in the caption.
• Overlays: compare series only when units and X definitions match; otherwise split them into panels.
• Trends and smoothing: a running average or polynomial trend is a visual summary. Record the window or degree, and move to a proper fit when the parameters matter.

For a model fitted to the data, the guide on how to fit a curve to data covers residuals and sensitivity checks.

Export the CSV graph for its destination

Choose the export from where the figure will end up:

Set the physical size first and the resolution second. Enlarging a small raster later blurs text and lines. Check labels, legend and line weights at the size the reader will see, and open the saved file rather than trusting the preview. The guide to exporting a vector PDF figure covers what stays vector, and making a multi-panel figure covers arranging several plots on one page.

Other ways to graph CSV data on a Mac

Numbers and Excel open a clean CSV and draw a chart in a few clicks. They are fine for a quick look, but they hide parser decisions, guess at data types and make it hard to see exactly which rows were plotted. Python with pandas and Matplotlib gives full control and a script you can rerun, at the cost of writing and maintaining code. A dedicated plotting app sits between the two: visible parser settings, named variables and repeatable figures without a notebook.

Whichever you choose, keep the source file, the parser settings and the figure together, so a reviewer's question can be answered by changing the analysis rather than editing an image.