In frequency encoding, each value in a categorical column is replaced with the total count or the frequency of the value. For example, let’s say a categorical column has 10 rows. The value “A” occurs 4 times, “B” occurs 3 times, “C” occurs 2 times and “D” occurs 1 time. So, we can replace A with 4, B with 3, C with 2, and D with 1. We can also divide these value counts with the total number of rows and replace A, B, C, and D with 0.4, 0.3, 0.2, and 0.1, respectively.

Let’s read the titanic dataset. Let’s encode the embark town of the dataset using frequency encoding. We can use the following Python code to replace each categorical value with the value counts of the value.

import seaborn

df = seaborn.load_dataset("titanic")

df.dropna(inplace=True)

value_counts = df["embark_town"].value_counts().to_dict()
print(value_counts)


df["embark_town"] = df["embark_town"].map(value_counts)
print(df.head())

Here, we are first getting the value counts of each categorical value of the embark town column and then, replacing the values with the value counts. We are using the pandas map function for this purpose.

The output of the above program will be:

{'Southampton': 115, 'Cherbourg': 65, 'Queenstown': 2}
    survived  pclass     sex   age  ...  deck  embark_town  alive  alone
1          1       1  female  38.0  ...     C           65    yes  False
3          1       1  female  35.0  ...     C          115    yes  False
6          0       1    male  54.0  ...     E          115     no   True
10         1       3  female   4.0  ...     G          115    yes  False
11         1       1  female  58.0  ...     C          115    yes   True

[5 rows x 15 columns]

We can also divide these value counts by the total number of rows of the dataset and use the frequency to replace each value of …