DataGraph alternative for Mac: when to switch and when to stay

DataGraph vs Autoplot at a glance

The fitting and data-model rows cut in DataGraph's favour, the scope row in Autoplot's, along with the packaged heavy-tail analysis described below. Which matters more depends on whether the next project is another 2D figure or a different kind of question.

Where DataGraph is the better choice

DataGraph is not a basic chart utility, and moving a working process has a cost. Stay with it when:

• The workflow is 2D and already validated. If a graph, its expressions and its fit have been checked, rebuilding them elsewhere adds work without improving the result.
• The model is a custom nonlinear function of X. DataGraph fits user-defined models with parameter bounds, weighted least squares and fit ranges, and extracts residuals. Autoplot's general-purpose route for that is SciPy code, written by you or by the assistant.
• Columns drive each other. Expression columns that update when the source changes are central to how DataGraph works.
• You export movies. DataGraph exports MP4 animations; Autoplot exports still images and PDFs.

For 3D work, the developer points DataGraph users, in a reply on its App Store listing, to its separate app ImageTank. That is worth checking before you change tools entirely.

Autoplot as a DataGraph alternative: beyond the 2D graph

3D surfaces and scatter plots

When height is part of the question, a 2D projection hides it. Autoplot renders surfaces, 3D scatter and line trajectories with camera and projection controls, and fits Gaussian or custom surfaces. Max Points decimates dense clouds with a deterministic stride. The 3D scatter plot guide covers when a third axis helps and when it only adds occlusion.

Heavy-tailed distributions as cards

Autoplot packages the methods a heavy-tail analysis needs: log-binned PDFs, a least-squares power-law and truncated power-law fit, an empirical CCDF with a tail fit through the powerlaw library, a Kolmogorov–Smirnov xmin scan, finite-size-scaling collapse and scaling relations of the form x ∼ yᵅ. Each runs locally as a tested Python implementation and writes its result back as variables. How to fit a power law to data explains which method to use.

A route to Python

For anything not built in, the assistant writes Python, runs it on your Mac inside a working folder and keeps the script. You see the code before it runs. A figure can also be exported as an editable Matplotlib script.

Move one DataGraph file to Autoplot

1. Keep the original DataGraph file and note the version that opens it.
2. Choose a project that needs something new, such as a 3D view or a tail fit. Moving a finished 2D graph proves little.
3. Export the source columns from DataGraph as delimited text, keeping the raw columns rather than only computed ones.
4. Import the file into Autoplot and check the parser preview: delimiter, header row, row count and missing values.
5. Recreate the derived columns as Autoplot variables and check a few values against DataGraph.
6. Run the new analysis, then compare any shared fit by its parameters and residuals, not by the overlay.
7. Keep the 2D graphs that already work in DataGraph if nothing is gained by moving them.

For the residual checks, see how to interpret residual plots.