Files
hpos-data/scripts/dataops39.py
Pritimay Sarkar 58f0fa0665 data ops
2023-11-08 14:03:31 +05:30

24 lines
871 B
Python

import pandas as pd
import os
import numpy as np
import matplotlib.pyplot as plt
curdir = os.getcwd()
path_delim = '/'
df1 = pd.read_excel(curdir + path_delim + "data/ABHA Aadhaar Number Issues.xlsx", sheet_name="Sheet1")
df2 = pd.read_excel(curdir + path_delim + "data/users_20_09_2023_06_17.xlsx", sheet_name="Sheet1")
print(df1)
df = df1.merge(df2, on="Sample ID", how="inner")
print(df.columns)
df = df[['S.No', 'Sample ID', 'Name', 'ABHA ID', 'abhaId', 'Aadhaar ID', 'aadharId',
'Age', 'Gender', 'Caste', 'caste', 'Category', 'Marital Status', 'Address',
'District', 'State', 'Pincode', 'Mobile Number', 'Date', 'Test Result',
'Blood Group']]
df_final = df
writer = pd.ExcelWriter(curdir + path_delim + "data/ABHA Aadhaar Number Issues-fix.xlsx", engine = 'openpyxl')
df_final.to_excel(writer, sheet_name = 'op', index=False)
writer.close()