Skip to main content

Pandas - CSV + df 예제

# %% [markdown]
# CSV + Dataframe 예제
# - 주피터 노트북에서도 파이썬 패키지 참조가 가능하다.(단, 로컬 가상환경)

# %%
import os
import pandas as pd
from file_io import save_df_to_csv, save_df_to_tsv, load_df_from_csv, load_df_from_tsv, add_columns_if_not_exists, clean_columns
import re
import time

print(os.getcwd())
FILE_PATH = "./output.csv"


# %%
# eg) save file to tsv
data = {
'이름': ['김철수', '이영희', '박민수'],
'나이': [25, 28, 30],
'성별': ['남', '여', '남']
}
df = pd.DataFrame(data)
save_df_to_tsv(df, "./output.tsv")
df

# %%
# eg) load file from tsv
df = load_df_from_tsv("./output.tsv")
df

# %%
# eg) add columns if not exists
df = load_df_from_tsv("./output.tsv")
add_columns_if_not_exists(df, ["userId","profile","summary"])
save_df_to_tsv(df, "./output.tsv")
df

# %%
# eg) add user id value
df = load_df_from_tsv("./output.tsv")
for i in range(len(df)):
df.loc[i, "userId"] = df.iloc[i]["이름"] + str(df.loc[i]["나이"]) + df.loc[i]["성별"] + str(i)
df.loc[i, "profile"] = df.iloc[i]["이름"] +"("+ str(df.loc[i]["나이"])+")"
df.loc[i, "summary"] = df.iloc[i]["이름"] + "(" + str(df.loc[i]["나이"]) + ")" + "(" + df.loc[i]["성별"] + ")"
save_df_to_tsv(df, "./output.tsv")
time.sleep(1)

# %%
# eg) clean only summary column
df = load_df_from_tsv("./output.tsv")
df = clean_columns(df, ["summary"])
save_df_to_tsv(df, "./output.tsv")
df