pandas

Question 1

Rename columns

Answer 1

df.rename(columns = {“oldName1”: “newName1”, “oldName2”:”newName2”})

Question 2

groupby

Answer 2

df.groupby([‘col1’, ‘col2’]).agg({‘col3’:’how’})

Question 3

replace NaN with zeros

Answer 3

df.fillna(0)

Question 4

create a new column for year-month

Answer 4

df[‘YM’] = df[‘Year’].astype(str) + ‘-‘ + df[‘Month’].astype(str)

Question 5

filter by multiple values in one column, and one specific value in another. (eg, 2015 + 2016 data for Germany)

Answer 5

df[df.col1.isin(list) & (df.col2 == ‘text’)]

Question 6

turn df into np_matrix

Answer 6

df.as_matrix()

Question 7

use index as a list

Answer 7

df.index.tolist()

Question 8

create a df

Answer 8

pd.DataFrame({‘colName1’:list1, ‘colName2’:list2})

Question 9

combine two lists into one df

Answer 9

pd.concat([list1, list2], axis = 1,keys= [‘colName1’, ‘colName2’])

Question 10

rename all columns to a certain list

Answer 10

df.columns = list

Question 11

get a total column for cross_tab

Answer 11

df[‘Total’] = df.sum(axis=1)

Question 12

get a total row for cross_tab

Answer 12

df.append(df.sum().rename(‘Total’))

Question 13

print names of columns

Answer 13

print(df.columns.values)

Question 14

count how many rows for each value in col1 (without using .agg)

Answer 14

df = df.groupby('col1').size()
df = df.reset_index(name='count')

Question 15

create a new column for running total

Answer 15

df[‘cumsum’] = df[‘col_to_sum’].cumsum()

Question 16

melt df (to prepare for side by side scatter plot) so that col1 is unique (goes on x-axis), col2 is the second level (different hue in plot), col3 is quantity

Answer 16

plot_data = pd.melt(df, id_vars=”col1”, var_name=”col2”, value_name=”col3”)

Question 17

mean/median/mode of col1

Answer 17

df.col1.mean() / df.col1.median() / df.col1.mode()