FitComparisonManager

This class handles the storage, display, and export of multiple fitting results for comparison. It is used from within a FitTab to enable model benchmarking and result visualization.

Role in the Application

  • Attached to each FitTab instance.
  • Activated via the toolbar ("Compare" action).
  • Stores all fit results (only 1D).

  • Provides a dock with:

  • A matplotlib plot showing all selected models

  • A table listing fit parameters and scores
  • A list to rename, enable/disable individual fits
  • A CSV export button

Comparison Flow

flowchart TD FTab[FitTab] FTab -->|calls| Comp[FitComparisonManager] Comp -->|store| Store[stored_fits] Comp -->|display| Dock[Comparison Dock] Dock --> PlotComp[Comparison Plot] Dock --> TableComp[Comparison Table]

Stored Data

Each entry in stored_fits is a dict containing:

{
    "formula": str,
    "method": str,  # "lmfit", "odr", "emcee"
    "params": lmfit.Parameters or dict,
    "aic": float,
    "bic": float,
    "rmse": float,
    "r_squared": float,
    "chi_square": float,
    "reduced_chi_square": float,
}

Supported Strategies

  • 1D Fit: Overlays different model predictions on a scatter plot of X, Y.
  • Fit per Y (multi-1D): Recomputes predictions for each Y slice, plots curves on X, Z.

Key Methods

Method Description
store_current_fit() Saves the current model result to internal list
toggle_comparison_mode() Shows/hides the comparison interface
create_comparison_ui() Builds dock with plot, list, table, and export button
redraw_comparison_plot() Redraws model predictions over data depending on selected fits
export_comparison_csv() Saves all stored results to a CSV file
reset() Clears all stored data and disables comparison mode

Dependencies

  • Requires fit_tab.plot_widget.result to be set.
  • Uses parse_formula() to dynamically recreate fit functions for plotting.
  • Compatible with both lmfit.ModelResult and custom dicts from emcee/odr.