Commit 119ef098 authored by Jan Kiene's avatar Jan Kiene
Browse files

add write out option for histograms to csv

parent 93d77e5f

scripts/create_histograms.py

+23 −3
Original line number Diff line number Diff line
@@ -37,17 +37,22 @@ def create_histograms( 
    display_only: bool,

    bins_for_measures=BINS_FOR_MEASURES,

    prefix="",

    write_out_histograms=False,

):

    formats = df["format"].unique()

    categories = df["category"].unique()



    if not display_only:

    if not display_only or write_out_histograms:

        output_folder.mkdir(exist_ok=True, parents=True)



    for measure in measures:

        measure_in_df = prefix + measure

        bins = bins_for_measures.get(measure, get_bins_for_diff(df[measure_in_df]))

        x = [f"{x}" for x in bins] + ["", "ERROR"]



        df_hist = pd.DataFrame(columns=["format", "category"] + x)

        hist_row_count = 0



        for fmt in formats:

            fig, ax = plt.subplots()

            ax.xaxis.set_major_formatter("{x:.1f}")
@@ -57,9 +62,9 @@ def create_histograms( 
                df_slice = df[data_mask]

                error_mask = df_slice["result"] == "ERROR"

                n_errors = np.sum(error_mask)

                df_hist = df_slice[np.logical_not(error_mask)]

                df_slice = df_slice[np.logical_not(error_mask)]



                counts, _ = np.histogram(df_hist[measure_in_df], bins)

                counts, _ = np.histogram(df_slice[measure_in_df], bins)



                data = np.concatenate([counts, [0], [n_errors], [0]])

                ax.bar(
@@ -74,6 +79,10 @@ def create_histograms( 
                )

                bottom += data



                hist_row = [fmt, cat] + list(counts) + [0] + [0, n_errors]

                df_hist.loc[hist_row_count] = hist_row

                hist_row_count += 1



            # Histogram layout

            ax.set_title(fmt)

            ax.legend(loc="best")
@@ -94,6 +103,11 @@ def create_histograms( 
                plt.savefig(image_path)

                plt.close(fig)



        if write_out_histograms:

            df_hist.to_csv(

                output_folder.joinpath(f"histogram_{measure}.csv"), index=False

            )



    if display_only:

        plt.show()
@@ -133,6 +147,11 @@ Use this for visualising diff scores.""", 
        default="",

        help="Common suffix to use when collecting measures from the input csv file",

    )

    parser.add_argument(

        "--write-out-histograms",

        action="store_true",

        help="Write out the histogram values to csv",

    )

    args = parser.parse_args()

    df = pd.read_csv(args.csv_report)
@@ -152,4 +171,5 @@ Use this for visualising diff scores.""", 
        args.display_only,

        bins_for_measures,

        args.prefix,

        args.write_out_histograms,

    )