Compare two Pandas DataFrames

compare_dfs.py

import pandas as pd
import numpy as np

def compare_two_dfs(input_df_1, input_df_2):
    
    # explicitly calling fillna with ""
    # as if you've used np.nan it has the 
    # property of nevery being able to be equals
    # i.e. `np.nan == np.nan` will always be False
    df_1, df_2 = input_df_1.copy().fillna(""), input_df_2.copy().fillna("")    
    
    ne_stacked = (df_1 != df_2).stack()
    changed = ne_stacked[ne_stacked]
    changed.index.names = ["id", "col"]
    difference_locations = np.where(df_1 != df_2)

    changed_from = df_1.values[difference_locations]

    changed_to = df_2.values[difference_locations]

    df = pd.DataFrame({"from": changed_from, "to": changed_to}, index=changed.index)
    return df

Using blank strings for fillna value probably isn't as performance conscious as some other known value.
But meh for now.