Get statistical data (median/mean/count/90%/10%) from multiple column combinations of a data frame

Question

Have a dataframe like below with 3 main variables, first variable has 2 unique values (True/false), second variable has 3 unique values (1/2/3), and third has 3 unique values (A/B/C), combining for 18 possible combinations.

Index	Var1	Var2	Var3	Value
0	TRUE	1	A	12
1	FALSE	2	B	93
2	TRUE	3	C	28
3	FALSE	1	A	23
4	TRUE	2	B	94
5	FALSE	3	C	12
6	TRUE	1	A	85
7	FALSE	2	B	23
8	TRUE	3	C	2

I want to cycle through these combinations and then print out a data frame similar to the one shown below for each combination (18 total dfs), filled with statistical data, which would most likely require a function but I'm not exactly sure the kind of logic needed for it.

(True/1/A)

Var1	Var2	Var3	Med	Mean	Count	90% quantile	10% quantile
TRUE	1	A	48.5	48.5	2	77.7	19.3

I tried grouping by which has worked well for combining everything, but didn't know med/mean/count/90% quantile and 10% quantiles well. I also tried querying to isolate each combination, but couldn't solve of the best way to iterate through the unique combos. Finally for quantiles I've tried .quantile().

Answer 1

Try:

df = df.groupby(["Var1", "Var2", "Var3"]).agg(
    Med=("Value", "median"),
    Mean=("Value", "mean"),
    Count=("Value", "count"),
    q90=("Value", lambda x: x.quantile(q=0.9)),
    q10=("Value", lambda x: x.quantile(q=0.1)),
)
df = df.reindex(
    pd.MultiIndex.from_product(
        [[True, False], [1, 2, 3], ["A", "B", "C"]], names=["Var1", "Var2", "Var3"]
    )
)

# print 18 dataframes:
for _, g in df.groupby(level=[0, 1, 2]):
    print(g)
    print("-" * 80)

Prints:

...

                  Med  Mean  Count   q90   q10
Var1  Var2 Var3                               
False 3    C     12.0  12.0    1.0  12.0  12.0
--------------------------------------------------------------------------------
                 Med  Mean  Count   q90   q10
Var1 Var2 Var3                               
True 1    A     48.5  48.5    2.0  77.7  19.3
--------------------------------------------------------------------------------
                Med  Mean  Count  q90  q10
Var1 Var2 Var3                            
True 1    B     NaN   NaN    NaN  NaN  NaN

...