smiuser/hpos-data
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

dedup

Browse Source
This commit is contained in:
Pritimay Sarkar
2023-09-19 18:03:24 +05:30
parent fcf0706881
commit 998abbee86
1 changed files with 7 additions and 2 deletions
Download Patch File Download Diff File Expand all files Collapse all files

9
scripts/combine2.py
View File

@@ -132,9 +132,14 @@ df_final.loc[df_final['Test Result'] == 'NBL', 'Test Result'] = 'Negative Border
df_final.loc[df_final['Gender'] == 'Male', 'Gender'] = 'MALE'
df_final.loc[df_final['Gender'] == 'Female', 'Gender'] = 'FEMALE'
print(df_final.groupby(["Gender"]).describe())
#df_final = df.drop_duplicates(subset=['Sample ID', "Test Time"], keep='last')
df_dedup = pd.read_excel(curdir + path_delim + "data/July_Sept_12_shared.xlsx", sheet_name="op")
print(df_dedup)
print(df_dedup.groupby(["Gender"]).describe())
writer = pd.ExcelWriter(curdir + path_delim + "data/July_Sept.xlsx", engine = 'openpyxl')
df_final.to_excel(writer, sheet_name = 'op', index=False)
writer.close()
writer.close()