satlas.utilities.plotting.generate_correlation_map

satlas.utilities.plotting.generate_correlation_map(f, x_data, y_data, method='chisquare_spectroscopic', filter=None, resolution_diag=20, resolution_map=15, fit_args=(), fit_kws={}, distance=5, npar=1)

Generates a correlation map for either the chisquare or the MLE method. On the diagonal, the chisquare or loglikelihood is drawn as a function of one fixed parameter. Refitting to the data each time gives the points on the line. A dashed line is drawn on these plots, with the intersection with the plots giving the correct confidence interval for the parameter. In solid lines, the interval estimated by the fitting routine is drawn. On the offdiagonal, two parameters are fixed and the model is again fitted to the data. The change in chisquare/loglikelihood is mapped to 1, 2 and 3 sigma contourmaps.

Parameters:

• f (BaseModel) – Instance of the model for which the contour map has to be generated.
• x_data (array_like or list of array_likes) – Data on the x-axis for the fit. Must be appropriate input for f.
• y_data (array_like or list of array_likes) – Data on the y-axis for the fit. Must be appropriate input for f.

Other Parameters:  

• method ({‘chisquare’, ‘chisquare_spectroscopic’, mle’}) – Chooses between generating the map for the chisquare routine or for the likelihood routine.
• filter (list of strings) – Only the parameters matching the names given in this list will be used to generate the maps.
• resolution_diag (int) – Number of points for the line plot on each diagonal.
• resolution_map (int) – Number of points along each dimension for the meshgrids.
• fit_kws (dictionary) – Dictionary of keywords to pass on to the fitting routine.
• npar (int) – Number of parameters for which simultaneous predictions need to be made. Influences the uncertainty estimates from the parabola.