Unnamed: 0  id     diet  pulse    time  kind
0           0   1  low fat     85   1 min  rest
1           1   1  low fat     85  15 min  rest
2           2   1  low fat     88  30 min  rest
3           3   2  low fat     90   1 min  rest
4           4   2  low fat     92  15 min  rest
no fat     45
low fat    45
Name: diet, dtype: int64

Now, we can use the following Python code to perform ordinal encoding on the diet column.

import seaborn
from sklearn.preprocessing import OrdinalEncoder

df = seaborn.load_dataset("exercise")
print(df.head())
print(df.diet.value_counts())

ordinal_encoder = OrdinalEncoder()
df[["diet"]] = ordinal_encoder.fit_transform(df[["diet"]])
print(df.diet.value_counts())

Here, we are using the OrdinalEncoder class from the sklearn.preprocessing module to perform ordinal encoding. The ordinal_encoder.fit_transform(df[[“diet”]]) function, learns about the ordinal data from the diet column of the dataset and then, it encodes the data and transforms it.

After the ordinal encoding is done, we can see the diet columns contain only values 0.0 and 1.0. So, the output of the above program will be:

   Unnamed: 0  id     diet  pulse    time  kind
0           0   1  low fat     85   1 min  rest
1           1   1  low fat     85  15 min  rest
2           2   1  low fat     88  30 min  rest
3           3   2  low fat     90   1 min  rest
4           4   2  low fat     92  15 min  rest
no fat     45
low fat    45
Name: diet, dtype: int64
0.0    45
1.0    45
Name: diet, dtype: int64

Please note that by default the desired data type of the output is numpy.float64. So, we get the numbers 0.0 and 1.0 as the encoded values of the diet column.