LEARN DATA SCIENCE ONLINE

Start Learning For Free - www.dataquest.io

Data Science Cheat Sheet

Pandas

KEY IMPORTS
Well use shorthand in this cheat sheet Import these to start
df - A pandas DataFrame object import pandas as pd
s - A pandas Series object import numpy as np

I M P O RT I N G DATA SELECTION ascending=[True,False]) - Sort values by col1 in

pd.read_csv(filename) - From a CSV file df[col] - Return column with label col as Series ascending order then col2 in descending order
pd.read_table(filename) - From a delimited text df[[col1, col2]] - Return Columns as a new df.groupby(col) - Return a groupby object for
file (like TSV) DataFrame values from one column
pd.read_excel(filename) - From an Excel file s.iloc[0] - selection by position df.groupby([col1,col2]) - Return a groupby
pd.read_sql(query, connection_object) - s.loc[0] - selection by index object values from multiple columns
Read from a SQL table/database df.iloc[0,:] - first row df.groupby(col1)[col2].mean() - Return the
pd.read_json(json_string) - Read from a JSON df.iloc[0,0] - first element of first column mean of the values in col2, grouped by the values
formatted string, URL or file. in col1 (mean can be replaced with almost any
pd.read_html(url) - Parses an html URL, string or DATA C L E A N I N G function from the statistics section)
file and extracts tables to a list of dataframes df.columns = ['a','b','c'] - Rename columns df.pivot_table(index=col1,values=
pd.read_clipboard() - Takes the contents of your pd.isnull() - Checks for null Values, Returns [col2,col3],aggfunc=max) - Create a pivot table
clipboard and passes it to read_table() Boolean Arrray that groups by col1 and calculates the mean of
pd.DataFrame(dict) - From a dict, keys for col- pd.notnull() - Opposite of s.isnull() col2 and col3
umns names, values for data as lists df.dropna() - Drop all rows that contain null df.groupby(col1).agg(np.mean) - find the
values average across all columns for every unique column
E X P O RT I N G DATA df.dropna(axis=1) - Drop all columns that con- 1 group
df.to_csv(filename) - Write to a CSV file tain null values data.apply(np.mean) - apply a function across
df.to_excel(filename) - Write to an Excel file df.dropna(axis=1,thresh=n) - Drop all rows each column
df.to_sql(table_name, connection_object) - have have less than n non null values data.apply(np.max, axis=1) - apply a function
Write to a SQL table df.fillna(x) - Replace all null values with x across each row
df.to_json(filename) - Write to a file in JSON s.fillna(s.mean()) - Replace all null values with
format the mean (mean can be replaced with almost any J O I N /C O M B I N E
df.to_html(filename) - Save as an HTML table function from the statistics section) df1.append(df2) - Add the rows in df1 to the end
df.to_clipboard() - Write to the clipboard s.astype(float) - Convert the datatype of the of df2 (columns should be identical)
series to float df.concat([df1, df2],axis=1) - Add the
C R E AT E T E ST O B J E C TS s.replace(1,'one') - Replace all values equal to columns in df1 to the end of df2 (rows should be
Useful for testing 1 with 'one' identical)
pd.DataFrame(np.random.rand(20,5)) - 5 col- s.replace([1,3],['one','three']) - Replace all df1.join(df2,on=col1,how='inner') - SQL-style
umns and 20 rows of random floats 1 with 'one' and 3 with 'three' join the columns in df1 with the columns on df2
pd.Series(my_list) - Create a series from an df.rename(columns=lambda x: x + 1) - mass where the rows for col have identical values. how
iterable my_list renaming of columns can be one of 'left', 'right', 'outer', 'inner'
df.index = pd.date_range('1900/1/30', df.rename(columns={'old_name': 'new_
periods=df.shape[0]) - Add a date index name'}) - selective renaming STAT I ST I C S
df.set_index('column_one') - change the index These can all be applied to a series as well.
V I E W I N G/ I N S P E C T I N G DATA df.rename(index=lambda x: x + 1) - mass df.describe() - Summary statistics for numerical
df.head(n) - First n rows of the DataFrame renaming of index columns
df.tail(n) - Last n rows of the DataFrame df.mean() - Return the mean of all columns
df.shape() - Number of rows and columns F I LT E R, S O R T, & G R O U P BY df.corr() - finds the correlation between columns
df.info() - Index, Datatype and Memory informa- df[df[col] > 0.5] - Rows where the col column in a DataFrame.
tion is greater than 0.5 df.count() - counts the number of non-null values
df.describe() - Summary statistics for numerical df[(df[col] > 0.5) & (df[col] < 0.7)] - in each DataFrame column.
columns Rows where 0.7 > col > 0.5 df.max() - finds the highest value in each column.
s.value_counts(dropna=False) - View unique df.sort_values(col1) - Sort values by col1 in df.min() - finds the lowest value in each column.
values and counts ascending order df.median() - finds the median of each column.
df.apply(pd.Series.value_counts) - Unique df.sort_values(col2,ascending=False) - Sort df.std() - finds the standard deviation of each
values and counts for all columns values by col2 in descending order column.
df.sort_values([col1,col2],

LEARN DATA SCIENCE ONLINE

Start Learning For Free - www.dataquest.io