frequency table ( cross table ), groupby and normalization in Python

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from scipy import stats

da = pd.read_csv("C:/Users/eli/Desktop/YtPruboBEemdqA7UJJ_tgg_63e179e3722f4ef783f58ff6e395feb7_nhanes_2015_2016.csv")

da["DMDEDUC2x"] = da.DMDEDUC2.replace({1: "<9", 2: "9-11", 3: "HS/GED", 4: "Some college/AA", 5: "College",
                                       7: "Refused", 9: "Don't know"})
da["DMDMARTLx"] = da.DMDMARTL.replace({1: "Married", 2: "Widowed", 3: "Divorced", 4: "Separated", 5: "Never married",
                                      6: "Living w/partner", 77: "Refused"})
db = da.loc[(da.DMDEDUC2x != "Don't know") & (da.DMDMARTLx != "Refused"), :]

# Now we can create a contingency table, counting the number of people in each cell defined by a combination of education and marital status.
x = pd.crosstab(db.DMDEDUC2x, da.DMDMARTLx)

# Normalize data
# A contingency table can be normalized in three ways -- we can make the rows sum to 1,
# the columns sum to 1, or the whole table sum to 1. Below we normalize within rows.
# This gives us the proportion of people in each educational attainment category who fall into each group of
# the marital status variable.
# Normalizing within the rows.
x.apply(lambda z: z/z.sum(), axis=1)
print(x)

# We can also normalize within the columns.
x.apply(lambda z: z/z.sum(), axis=0)

# # The following line does these steps, reading the code from left to right:
# 1 Group the data by every combination of gender, education, and marital status
# 2 Count the number of people in each cell using the 'size' method
# 3 Pivot the marital status results into the columns (using unstack)
# 4 Fill any empty cells with 0
# 5 Normalize the data by row
b = da.groupby(["RIAGENDRx", "DMDEDUC2x", "DMDMARTLx"]).size().unstack().fillna(0).apply(lambda x: x/x.sum(), axis=1)
print(b.loc[:, ["Married"]].unstack())