Apply

Single Col

def funcName(num):
	return num + 1

ad['colName'].apply(funcName)

Multi Col

def funcName(col1, col2):
	return col1 + col2

df[['col1', 'col2']].apply(lambda df : funcName(df['col1'], df['col2']), axis=1)
# or
np.vectorize(funcName)(df['col1'],df['col2'])

Description

df.describe()

df['col'].max()
df['cdl'].idxmax()

df.corr() # describes the corrilation between cols

df['col'].value_counts() # counts num of each value

df['col'].unique() # what are the unique values
df['col'].nunique() # how many are there


df['col'].replace(['old', 'old1'], ['new', 'new1'])

# does the same as replace
mymap = {'old':'new', 'old1':'new1'}
df['col'].map(mymap)

df.duplicated() # returns a boolean_list with true on first dup
df.drop_duplicates() 

df['col'].between(lowIndex,upperIndex,inclusive=True)
# returns boolean_list 

df.nlargest(2,'colName') # returns the 2 largest values
df.nsmallest(2,'colName') 

df.sample(5) # 5 random rows
df.sample(frac=0.1) # random 10% of rows

Sorting

df.sort_values(['colName', 'colName2'], ascending=True)

df['col'].max()
df['col'].idxmax()

String Methods

df.str.METHOD() # does method on everything

df.str.isdigit() # boolean series

df.str.split(',') # creates an array with elements split on ','
# expand=True this turns the array into cols

#cleaning
df.str.replace(';', '')
df.str.strip() # removes white space

Time Methods

from datetime import datetime

date = datetime(year, month, day, hours, mins, secs) # all vars are ints

# note in df probs date stored as a string
# this converts most string formats
df['date'] = pd.to_datetime(df['date'])
# dayfirst=True, so you can specify if the day is first in the date, 
# although it does a good job of guessing, but use the above to be safe
# format='%d--%b--%Y'