Function plot_summary

experiments/exp_longitudinal_accuracy.py:93–131

(resultsdir, neuralnet, savepath=None)

        print(f"writing {csvfile}")

def plot_summary(resultsdir, neuralnet, savepath=None):

    runs = os.listdir(resultsdir)
    runs = [r for r in runs if os.path.isdir(os.path.join(resultsdir,r))]

    colors = ['#e41a1c',  # Red
              '#377eb8',  # Blue
              '#4daf4a',  # Green
              '#984ea3']  # Purple

    runs = [r for r in runs if neuralnet in r]
    legend_names = [n.replace(f"-{neuralnet}", "") for n in runs]

    N = len(runs)

    fig, ax = plt.subplots()

    for o, run, color in zip(np.linspace(-N/2, N/2,N), runs, colors):
        file = np.load(os.path.join(resultsdir, run, "histogram.npz"))
        hist_accuracy = file["hist_accuracy"]
        bin_width = file["bin_width"]

        bin_edges = file["bin_edges"] #np.linspace(-90,90, 10)
        heights = np.diff(bin_edges) * 0.9 / N

        offset = o * heights[0]/1.5

        ax.barh(bin_edges[:-1] + bin_width / 2 + offset, hist_accuracy,
                height=heights, align='center', color=color)

    ax.legend(legend_names, ncols=1, loc="lower left")

    ax.set_xlabel("accuracy")
    ax.set_ylabel("latitude")
    ax.set_yticks(bin_edges[:-1] + bin_width / 2)

    if savepath is not None:
        print(f"writing {savepath}")
        fig.savefig(savepath, bbox_inches="tight", pad_inches=0, transparent=True)

if __name__ == '__main__':
    main()