And here is a sample implementation:

unique_flairs = list(df['Flair'].unique())
flair_df = pd.DataFrame()

# iterate over unique dates
for date in df['Submission_Date'].unique():
    date_subset = df.loc[df['Submission_Date'] == date]
    # for flairs in each date, get counts of values
    counts = date_subset['Flair'].value_counts()
    # get shares
    shares = counts / len(date_subset)
    # transpose series for appending
    shares_df = pd.DataFrame(shares).transpose()
    shares_df['Submission_Date'] = date
    for flair in [x for x in unique_flairs if x not in shares_df.columns]:
        shares_df[flair] = np.nan
    # append shares per date
    flair_df = pd.concat([flair_df, shares_df])
    flair_df = flair_df.reset_index(drop=True)

# rearrange columns
flair_df = flair_df[['Submission_Date']   unique_flairs]
display(flair_df)

Output:

CodePudding user response：

You can use pivot table to achieve this result. I made up the input data frame with some random values.

data = pandas.DataFrame.from_dict({
    "Submission_Date": [
        datetime.date(2021, 1, 1),
        datetime.date(2021, 1, 1),
        datetime.date(2021, 1, 2),
        datetime.date(2021, 1, 2),
        datetime.date(2021, 1, 3),
        datetime.date(2021, 1, 3),
        datetime.date(2021, 1, 3),
        datetime.date(2021, 1, 4),
    ],
    "Flair": ["Discussion", "Due Diligence", "Due Diligence", "Discussion", "Discussion",  "Hedge Fund Tears", "News", "News"],
})
data["Flair1"] = data.Flair.values # copy to another column to assist pivot
res = pandas.pivot_table(
    data,
    index=["Submission_Date"],
    values=['Flair'],
    columns=['Flair1'],
    aggfunc='count',
    fill_value=0

)
res = pandas.DataFrame(res.to_records())
res.columns = [col.replace("('Flair', ", '').replace(")", '') for col in res.columns]
res['Total'] = res.astype({col:float for col in res.columns if col != "Submission_Date"}).sum(numeric_only=True, axis=1) # find Total
res[[col for col in res.columns if col != "Submission_Date"]] = res[[col for col in res.columns if col != "Submission_Date"]].div(res.Total, axis=0) # divide by Total
res = res.drop(columns=['Total']) # drop Total
print(res)