Pandas Cheat Sheet

**import** *pandas* **as** *pd*

**# Method 1**
df1 = pd.DataFrame({
    *'name'*: ['John Smith', 'Jane Doe'],
    *'address'*: ['13 Main St.', '46 Maple Ave.'],
    *'age'*: [34, 28]
})

**# Method 2**
df2 = pd.DataFrame([
    ['John Smith', '123 Main St.', 34],
    ['Jane Doe', '456 Maple Ave.', 28],
    ['Joe Schmo', '9 Broadway', 51]
    ],
    **columns**=[*'name', 'address', 'age'*])

**# Load a CSV File in to a DataFrame**
df = pd.read_csv('my-csv-file.csv')

**# Saving DataFrame to a CSV File**
df.to_csv('new-csv-file.csv')

**# Load DataFrame in Chunks (For large Datasets)**
# Initialize reader object: urb_pop_reader
urb_pop_reader = pd.read_csv('ind_pop_data.csv', chunksize=1000)

# Get the first DataFrame chunk: df_urb_pop
df_urb_pop = next(urb_pop_reader)

nums = np.array(range(1, 11))
-> [ 1 2 3 4 5 6 7 8 9 10]

nums = nums.reshape(-1, 1)
-> [ [1],
[2],
[3],
[4],
[5],
[6],
[7],
[8],
[9],
[10]]

df = pd.DataFrame([
  ['January', 100, 100, 23, 100],
  ['February', 51, 45, 145, 45],
  ['March', 81, 96, 65, 96],
  ['April', 80, 80, 54, 180],
  ['May', 51, 54, 54, 154],
  ['June', 112, 109, 79, 129]],

  columns=['month', 'east', 'north', 'south', 'west']
)

**# Select one Column**
**clinic_north = df.north**
--> Reshape values for Scikit learn: clinic_north.**values.reshape(-1, 1)**

**# Select multiple Columns**
**clinic_north_south** = df[['north', 'south']]

**# Select one Row**
march = df.iloc[2]


**# Select multiple Rows**
jan_feb_march = df.iloc[:3]
feb_march_april = df.iloc[1:4]
may_june = df.iloc[-2:]


**# Select Rows with Logic**
january = df[df.month **==** 'January']
**-> <, >, <=, >=, !=, ==**

march_april = df[(df.month == 'March') | (df.month == 'April')]
**-> &, |**

january_february_march = df[df**.month.isin**(['January', 'February', 'March'])]
**-> column_name.isin([" ", " "])**

df = pd.DataFrame([
  [1, '3 inch screw', 0.5, 0.75],
  [2, '2 inch nail', 0.10, 0.25],
  [3, 'hammer', 3.00, 5.50],
  [4, 'screwdriver', 2.50, 3.00]
],
  columns=['Product ID', 'Description', 'Cost to Manufacture', 'Price']
)

**# Add a Column with specified row-values**
df['Sold in Bulk?'] = ['Yes', 'Yes', 'No', 'No']

**# Add a Column with same value in every row**
df['Is taxed?'] = 'Yes'

**# Add a Column with calculation**
df['Revenue'] = df['Price'] - df['Cost to Manufacture']

df = pd.DataFrame([
  ['JOHN SMITH', 'john.smith@gmail.com'],
  ['Jane Doe', 'jdoe@yahoo.com'],
  ['joe schmo', 'joeschmo@hotmail.com']
],
columns=['Name', 'Email'])

**# Changing a column with an Operation**
df['Name'] = df.Name.**apply(lower)**
*-> lower, upper*

**# Perform a lambda Operation on a Column**
get_last_name = **lambda x: x.split(" ")[-1]**
df['last_name'] = df.Name.**apply(get_last_name)**

df = pd.DataFrame([
  ["Apple", 1.00, "No"],
  ["Milk", 4.20, "No"],
  ["Paper Towels", 5.00, "Yes"],
  ["Light Bulbs", 3.75, "Yes"],
],
  columns=["Item", "Price", "Is taxed?"])


**# Lambda Function**
df['Price with Tax'] = df.apply(lambda row:
     row['Price'] * 1.075
     if row['Is taxed?'] == 'Yes'
     else row['Price'],
     **axis=1**
)

**# Method 1**
df**.columns** = ['NewName_1', 'NewName_2, 'NewName_3', '...']

**# Method 2**
df**.rename**(columns={
    'OldName_1': 'NewName_1',
    'OldName_2': 'NewName_2'
}, **inplace=True**)

**# Group By**
grouped = df.**groupby(['col1', 'col2'])**.col3
.*measurement()*.**reset_index()**
*# -> group by column1 and column2, calculate values of column3*

**# Percentile**
high_earners = df.groupby('category').wage
    .**apply(lambda** x: np.percentile(x, 75)**)**
    .**reset_index()**
*# np.percentile can calculate any percentile over an array of values*

orders = pd.read_csv('orders.csv')

shoe_counts = orders.
groupby(['shoe_type', 'shoe_color']).
id.count().reset_index()

shoe_counts_pivot = shoe_counts.pivot(
**index** = 'shoe_type',
**columns** = 'shoe_color',
**values** = 'id').reset_index()

sales = pd.read_csv('sales.csv')
targets = pd.read_csv('targets.csv')
men_women = pd.read_csv('men_women_sales.csv')

**# Method 1**
sales_targets = pd**.merge**(sales, targets, **how=**" ")
*# how: "inner"(default), "outer", "left", "right"*

**#Method 2 (Method Chaining)**
all_data = sales**.merge**(targets)**.merge**(men_women)

orders = pd.read_csv('orders.csv')
products = pd.read_csv('products.csv')

**# Method 1: Rename Columns**
orders_products = pd**.merge**(orders, products**.rename**(columns={'id':'product_id'}), **how=" "**)
**.reset_index()**
*# how: "inner"(default), "outer", "left", "right"*

**# Method 2: **
orders_products = 
pd**.merge**(orders, products,
                **left_on**="product_id",
                **right_on**="id",
                **suffixes**=["_orders","_products"])

bakery = pd.read_csv('bakery.csv')
ice_cream = pd.read_csv('ice_cream.csv')

menu = pd**.concat**([bakery, ice_cream])

# Test if country is of type object
**assert gapminder.country.dtypes == np.object**

# Test if year is of type int64
**assert gapminder.year.dtypes == np.int64**

# Test if life_expectancy is of type float64
**assert gapminder.life_expectancy.dtypes == np.float64**

# Assert that country does not contain any missing values
**assert pd.notnull(gapminder.country).all()**

# Assert that year does not contain any missing values
**assert pd.notnull(gapminder.year).all()**