Indices order of a resulting object returned from value_counts is differ from pandas

[YarShev]

Describe the problem

Indices order of a resulting object returned from values_counts and Series.values_counts is differ from pandas implementation. An order of indices is random in pandas, whereas we implemented the logic that sorts both values and indices.

Source code / logs

import modin.pandas as pd
import pandas
import numpy as np
NROWS = 2 ** 8
RAND_LOW = 0
RAND_HIGH = 100
random_state = np.random.RandomState(seed=42)
data = random_state.randint(RAND_LOW, RAND_HIGH, size=(NROWS))
ms = pd.Series(data)
ps = pandas.Series(data)
mr = ms.value_counts()
pr = ps.value_counts()
mr[:10] # sorted (by descending) both values and indices for same values
61    12
1      8
87     6
43     6
14     6
89     5
88     5
59     5
58     5
2      5
ps[:10] # sorted (by descending) values but not sorted indices for same values
61    12
1      8
43     6
14     6
87     6
88     5
2      5
59     5
58     5
89     5

[YarShev]

I think we can improve execution time for Series.value_counts removing the logic related to sorting indices for equal values. That logic can be kept for tests for now in order to Modin and pandas results are equal.

[YarShev]

The output for the scenario above after removing the sorting logic (#2454).

mr[:10]
61    12
1      8
43     6
14     6
87     6
59     5
2      5
88     5
89     5
58     5