segmentation_by_natural_partitioning.py

​​def segmentation_by_natural_partitioning(s):
    # calculate 5th and 95th percentiles.
    s_as_array = np.array(s)
    fith_percentile = np.percentile(s_as_array, 5)
    nienty_fith_percentile = np.percentile(s_as_array, 95)
    print(f"range [{s['A2'].max()},{s['A2'].min()}]")
    print()
    print("*****************************")
    print(f'fith_percentile {fith_percentile}')
    print(f'nienty_fith_percentile {nienty_fith_percentile}')
    print("*****************************")
   
    # sort the data.
    s['A2'] = s['A2'].sort_values()

    n = s['A2'].count()
    print("*****************************")
    print(f'Total number of records {n}')
    print("*****************************")
    # keep the values from floor(n*0.05) to floor(n*0.95)
    print("******************************")
    print("Dataset attribute A2")
    print(s['A2'])
    print("******************************")
    f1 = fith_percentile # np.math.floor(n*0.05)
    f2 = nienty_fith_percentile # np.math.floor(n*0.95)
    print("*****************************")
    print(f'fith_percentile {f1}')
    print(f'nienty_fith_percentile {f2}')
    print("*****************************")
    s = s[s['A2'] > f1]
    s = s[s['A2'] < f2]

    print(f"range after cleaning [{s['A2'].max()},{s['A2'].min()}]")

    maximum = ceil(s['A2'].max())
    minimum = floor(s['A2'].min())

    print(f"[maximum,minimum] [{minimum},{maximum}]")
    print(f"subtract {most_significant_while_floordiv(maximum)} - {most_significant_while_floordiv(minimum)}")
    numberOfGaps =  most_significant_while_floordiv(maximum) - most_significant_while_floordiv(minimum)
    print(f"numberOfGaps {numberOfGaps}")
    print(f"The number of values that cover the range are [{minimum},{maximum}]")

    return s