Function prediction_hist

climart/utils/plotting.py:317–362

(preds: dict, TOA=False, surface=False,
                    title="", show=True, figsize=(16, 12), axes=None,
                    label="", **kwargs)

def prediction_hist(preds: dict, TOA=False, surface=False,
                    title="", show=True, figsize=(16, 12), axes=None,
                    label="", **kwargs):
    n_vars = len(preds.keys())
    n_cols = 3 if TOA and surface else 2 if (TOA or surface) else 1

    surface_ax = 1
    TOA_ax = 2 if surface else 1

    if axes is None:
        fig, axs = plt.subplots(n_vars, n_cols, figsize=figsize)
        fig.suptitle("Prediction magnitudes" if title == "" else title)
        axs[0, 0].set_title('Mean')

        if surface:
            axs[0, surface_ax].set_title('Surface')
        if TOA:
            axs[0, TOA_ax].set_title('TOA')
    else:
        axs = axes

    def set_bar_colors(patches, upto=5):
        return
        jet = plt.get_cmap('jet', len(patches))
        for i in range(len(patches)):
            if i > upto:
                return
            patches[i].set_facecolor(jet(i * 10))

    for (var_name, var_preds), ax_row in zip(preds.items(), axs):
        # n_samples, n_levels = var_preds.shape
        N, bins, patches = ax_row[0].hist(np.mean(var_preds, axis=1), label=label, **kwargs)
        set_bar_colors(patches)
        ax_row[0].set_ylabel(f"{var_name.upper()}", fontsize=20)
        if surface:
            N, bins, patches = ax_row[surface_ax].hist(var_preds[:, -1], label=label, **kwargs)
            set_bar_colors(patches)
        if TOA:
            N, bins, patches = ax_row[TOA_ax].hist(var_preds[:, 0], label=label, **kwargs)
            set_bar_colors(patches)

    axs[0, 0].legend()
    if show:
        plt.show()

    return axs


def prediction_bars(preds: dict, bins, TOA=False, surface=False,