whylogs_pandas_dataframe_profiling_with_summary_statistics.py

1import whylogs as why
2import pandas as pd
3
4# 1. Create a simple dataset
5data = {
6    "column_1": [1, 2, 3, 4, 5],
7    "column_2": ["a", "b", "c", "d", "e"],
8    "column_3": [1.0, 2.0, 3.0, 4.0, 5.0]
9}
10df = pd.DataFrame(data)
11
12# 2. Log the data to create a profile
13results = why.log(df)
14
15# 3. Access the profile summary as a Pandas DataFrame
16profile_view = results.view()
17summary_df = profile_view.to_pandas()
18
19# 4. Print the summary
20print(summary_df)