Allow for cells where there are 0 outcomes in source data.

[jameson-c]

In cases where binary flags are being imputed, we need to allow for imputer.gen_analysis_file() to save summaries using columns with NaN results. This occurs if a cell has no instances of a '1' binary flag in donor data.

The following error is a clue that this is happening when you run imputer.gen_analysis_file():
TypeError: NAN/INF not supported in write_number() without 'nan_inf_to_errors' Workbook() option

[jameson-c]

Some code to troubleshoot + identify where the save process breaks down if working + don't have time to update package:

from xlsxwriter import Workbook
import os

results = imputer.summarize_cells()

full_path = os.path.join('YOUR PATH', 
                         f'hot_deck_stats.xlsx')

row = 1
col = 0

# Iterate through each cell's data
with Workbook(full_path) as wb:  
    ws = wb.add_worksheet('Summary')
    for key, df in results.items():
        ws.write(row-1, col, key)

        print(row, col, key)

        # Write table to excel  
        # WILL FAIL FOR CELL WHERE NO BINARY FLAG EXISTS IN SOURCE
        df.write_excel(workbook = wb, 
                           worksheet = ws,
                            position = (row, col),
                           table_style="Table Style Light 1",
                           autofit = True)
            
        # 2 row gap between each DF's results
        row = row + df.shape[0] + 3
print(f"Cell data written to hot_deck_stats.xlsx")