ENH: add value_counts_with_normalization

[dont-4get-me-men]

Feature Type

• Adding new functionality to pandas

• Changing existing functionality in pandas

• Removing existing functionality in pandas

Problem Description

I wish i could write one function which will create view for normalizied data and non-normalized data while calliing thi function.

Feature Description

i see this function in this way

def value_counts_with_normalization(self, subset=None,sort=True, ascending=False, dropna=True,percentage = False):
    ----
    non_normalized_df = self.value_counts(subset = subset
                 ,sort = True,ascending=ascending
                ,dropna = dropna, normalization = False).reset_index()
   
    normalized_df = self.value_counts(subset = subset,sort = True,ascending=ascending,dropna = dropna, normalization = False).reset_index()
    multiplier = 100 if percentage else 1
    normalized_df['proportion'] = normalized_df['proportion']*multiplier
return non_normalized_df.merge(normalized_df , on = subset)

Alternative Solutions

I couln't find alternative solution

Additional Context

Example of output

[dont-4get-me-men]

Also. If it will be okay to add this i would like to write this code by myself)

[jbrockmendel]

this is effectively the same as

left = ser.value_counts()
right = ser.value_counts(normalize=True)
result = pd.concat([left.to_frame("count"), right.to_frame("proportion")])

right? i dont think we want to add new API for this.

[simonjayhawkins]

I agree that introducing a new API for this purpose isn't necessary.

The pandas API already offers rich, composable tools for data wrangling, and the proposed output may not generalize well across use cases. Adding a dedicated function risks future enhancements, keyword proliferation, and ultimately API creep.

As @jbrockmendel demonstrated, the desired output can be achieved succinctly using existing methods.

This approach empowers users to tailor the output to their specific needs without constraining them to a fixed format.

Moreover, pandas provides multiple pathways to achieve similar results, e,g, s.value_counts().to_frame().assign(proportion=s.value_counts(normalize=True)) depending on the user's workflow and integration needs. A new API here might inadvertently reduce flexibility rather than enhance it.