4a24e4e36ee175aef54b92eb03e08a2be7811a96,recordlinkage/datasets/febrl.py,,_febrl_links,#Any#,27

Before Change

def _febrl_links(df):
    Get the links of a FEBRL dataset.

    df_empty = df[[]].reset_index()
    df_empty["key"] = df_empty["rec_id"].str. \
        extract(r"rec-(\d+)", expand=True)[0]

    // split the dataframe (org and dup)
    org_bool = df_empty["rec_id"].str.endswith("org")

    // merge the two frame and make MultiIndex.
    pairs = df_empty[org_bool].merge(df_empty[~org_bool], on="key")
    pairs_mi = pairs.set_index(["rec_id_x", "rec_id_y"]).index
    pairs_mi.names = [None, None]

    return pairs_mi

After Change

    index_int = numpy.arange(len(df))

    df_helper = pandas.DataFrame({
        "key": keys,
        "index": index_int
    })

    // merge the two frame and make MultiIndex.
    pairs_df = df_helper.merge(
        df_helper, on="key"
    )[["index_x", "index_y"]]
    pairs_df = pairs_df[pairs_df["index_x"] > pairs_df["index_y"]]

    return pandas.MultiIndex(
        levels=[df.index.values, df.index.values],

Instances