In this note, a general dataframe is called df (type pandas.core.frame.DataFrame), a general series is call s (type pandas.core.series.Series).

Import library

import pandas as pd
import numpy as np # import numpy if necessary

Read/Write .csv file

# READ
df = pd.read_csv('filename.csv', sep=';') # default sep=','

# if 1st col contains 0,1,2,...
df = pd.read_csv('filename.csv', index_col=1)

# with datetime info
df = pd.read_csv(PATH_DATA_FOLDER+"raw_data.csv",
                  parse_dates=['timestamp'],
                  infer_datetime_format=True,
                  cache_dates=True)

# WRITE
df.to_csv(path, index=False) # don't incldue index

Create a dataframe

# FROM A LIST
pd.DataFrame(a_list, colummns=['col_name'])

# FROM A DICTIONARY
names = ['John', 'Thi', 'Bi', 'Beo', 'Chang']
ages =  [10, 20, 21, 18, 11]
marks = [8, 9, 10, 6, 8]
city = ['Ben Tre', 'Paris', 'Ho Chi Minh Ville', 'New York', 'DC']

my_dict = {'Name':names, 'Ages':ages, 'Marks':marks, 'Place': city}
students = pd.DataFrame(my_dict)
Name Ages Marks Place
0 John 10 8 Ben Tre
1 Thi 20 9 Paris
2 Bi 21 10 Ho Chi Minh Ville
3 Beo 18 6 New York
4 Chang 11 8 DC

Adding

# a column
df['new_col] = [new_values]

# a row
df.loc['new_index'] = [new_value]

# add a new col based on another's values
df_im = df0.copy()[['col']]
df_im['status'] = df0['col'].apply(lambda row: 1 if row>=80 else 0)

Shuffle rows

# shuffle all rows and reset the index
df_new = df.sample(frac=1).reset_index(drop=True)

Sorting

df.sort_values(by='col1', ascending=False)

Select rows/columns/item(s)

👉 Indexing and selecting data — pandas 1.1.2 documentation