如何使用 Altair 将 % 添加到 Pandas pivot 表中

Question

I am working on a survey and the data looks like this:

ID    Q1    Q2    Q3    Gender    Age    Dept
001   Y      N    Y      F         22     IT
002   N      Y    Y      M         35     HR
003   Y      N    N      F         20     IT
004   Y      N    Y      M         54     OPRE
005   Y      N    Y      M         42     OPRE

So I created a pivot table like this:

Q1    #Respondents      %Res
Y        4               80
N        1               20

If I would like to slice it by Gender, then it should be like:

Q1      #Res        %Rep
       M    F      M    F
Y      2    2      50   50
N      1    0      100   0

And if I want this to be applied to all the questions, I'd like to use Altiar which enables me to choose the question so that I don't need to execute the codes all the time. So far, I only know how to create simple table by:

Q1 = pd.pivot_table(df,values = ['ID'], 
                index = ["Q1"], 
                aggfunc ={'ID': 'count', })
Q1['%Respondents'] = (Q1['ID']/Q1['ID'].sum())*100
Q1

I don't know how to break it by gender and apply Altair. Please let me know if you could help! Thanks!

Answer 1

To break by gender. Just utilized it as index, and then unstack it

Q1 = pd.pivot_table(df,values = ['ID'], 
                index = ["Q1","Gender"], 
                aggfunc ={'ID': 'count', }).unstack(level = 0)

Unfurtunataly cant help you with you graph

Answer 2

IIUC, you can pivot_table then add the result of division as a new foo column %Respondents

out = df.pivot_table(index='Q1', columns=['Gender'], values=['ID'], aggfunc='count', fill_value=0)
out = (out.join(out[['ID']].div(out['ID'].sum(axis=1).values, axis=0)
                .mul(100)
                .rename(columns={'ID':'%Respondents'})))

print(out)

       ID    %Respondents
Gender  F  M            F      M
Q1
N       0  1          0.0  100.0
Y       2  2         50.0   50.0

Answer 3

To keep the ball rolling, here how you could add % into altair charts.

• Step 1. Convert wide dataframe to a long one with melt. I made a simple bar chart with counts for Q1 - without percentage % yet.

df=pd.DataFrame({'ID':[1, 2, 3, 4, 5],
'Q1':['Y', 'N', 'Y', 'Y', 'Y'],
'Q2':['N', 'Y', 'N', 'N', 'N'],
'Q3':['Y', 'Y', 'N', 'Y', 'Y'],
'Gender':['F', 'M', 'F', 'M', 'M'],
'Age':[22, 35, 20, 54, 42],
'Dept':['IT', 'HR', 'IT', 'OPRE', 'OPRE']}).melt(id_vars=['ID', 'Gender', "Age", "Dept"])

alt.Chart(df.query('variable=="Q1"'), title="Q1").mark_bar().encode(
    x=alt.X('value:N', title="", axis=alt.Axis(labelAngle=0)),
    y=alt.Y('count()', axis=alt.Axis(tickMinStep=1)),
    color="Gender",
    column=alt.Column("Gender", title="")
).properties(width=100, height =150)

• Step 2: One of possible scenarios - how to calculate %:

qs=df.variable.unique().tolist()
charts=[]

for q in qs:
    chart=alt.Chart(df.loc[df['variable']==q], title=q).transform_joinaggregate(
        total='count(*)'
        ).transform_calculate(
        pct='1 / datum.total').mark_bar().encode(
            y=alt.Y('value:N', title=""),
            x=alt.X('sum(pct):Q', axis=alt.Axis(format='%'), title=""),
            tooltip=[alt.Tooltip('sum(pct):Q',format=".1%")],
            color="Gender"
        ).properties(width=200)
    charts.append(chart)

alt.hconcat(*charts).configure_scale(
        bandPaddingInner=0.03
    )