Home TechnologyCoding Data Science: Python and Pandas cheatsheets and functions.

Data Science: Python and Pandas cheatsheets and functions.

The Pandas DataFrame for Data Science and Data Analysis.

by Ivan
python pandas data science cheat sheet python cheatsheet

Python is one of the most widely used language for Data Analysis and Data Science and the pandas library is no exception. Python is easy to learn, has a great online community of learners and instructors, and has some really powerful data-centric libraries. Pandas is one of the most important libraries in Python for Data Analysis, and Data Science.

In this story, I will present the 13 most significant Pandas functions and methods that are essential for Data Analysis and Data Science to know, and will share with you some wonderfully designed cheat sheets for you to reference in your journey.

Download the most comprehensive Python Pandas Cheat Sheet here

Python-Pandas-Cheat-Sheets-1-8-v1.0.2.pdf
  1. The most comprehensive cheat sheet, from the Pandas team:
    https://pandas.pydata.org/Pandas_Cheat_Sheet.pdf
    It covers the vast majority of what most pandas users will ever need to do to a DataFrame.
  2. A compact, yet complete cheat sheet, from datacamp:
    http://datacamp-community-prod.s3.amazonaws.com/dbed353d-2757-4617-8206-8767ab379ab3
    Data Camp is an online platform that teaches Data Science with videos and coding exercises. They have made cheat sheets on a bunch of the most popular Python libraries, which you can also check out here.
  3. Cheatsheet for statistical analysis:
    https://www.webpages.uidaho.edu/~stevel/504/pandas%20dataframe%20notes.pdf
    This was put together by a student at the University of Idaho a few years ago. It might seem a little dry, but it shines with its complete section on time series and statistics. There are methods for calculating covariance, correlation, and regression here. So, if you are using pandas for some advanced statistics or any kind of scientific work, this is going to be your cheat sheet.

Now, for the functions:

1. read_csv()

read_csv() function helps read a comma-separated values (csv) file into a Pandas DataFrame. All you need to do is mention the path of the file you want it to read. It can also read files separated by delimiters other than comma, like | or tab. More details here.

data_1 = pd.read_csv(r'C:UsersABCDesktopblog_dataset.csv')

The data has been read from the data source into the Pandas DataFrame. You will have to change the path of the file you want to read. You can download the dataset used in the blog.

to_csv() function works exactly opposite of read_csv(). It helps to write data contained in a Pandas DataFrame or Series to a csv file. You can read more about to_csv() here. read_csv() and to_csv() are one of the most used functions in Pandas because they are used while reading data from a data source, and are very important to know.

2. head()

head(n) is used to return the first n rows of a dataset. By default, df.head() will return the first 5 rows of the DataFrame. If you want more/less number of rows, you can specify n as an integer.

data_1.head(6)

Output:

NameAgeCityStateDOBGenderCity tempSalary
0Alam29IndoreMadhya Pradesh20-11-1991Male35.550000
1Rohit23New DelhiDelhi19-09-1997Male39.085000
2Bimla35RohtakHaryana09-01-1985Female39.720000
3Rahul25KolkataWest Bengal19-09-1995Male36.540000
4Chaman32ChennaiTamil Nadu12-03-1988Male41.165000
5Vivek38GurugramHaryana22-06-1982Male38.935000

The first 6 rows (indexed 0 to 5) are returned as output as per expectation.

tail() is similar to head(), and returns the bottom n rows of a dataset. head() and tail() help you get a quick glance at your dataset, and check if data has been read into the DataFrame properly.

3. describe()

describe() is used to generate descriptive statistics of the data in a Pandas DataFrame or Series. It summarizes central tendency and dispersion of the dataset. describe() helps in getting a quick overview of the dataset. More details about describe() can be found here.

data_1.describe()

Output:

AgeCity tempSalary
count9.0000008.0000009.000000
mean32.00000038.57500044444.444444
std5.8949131.77180321360.659582
min23.00000035.50000018000.000000
25%29.00000038.30000035000.000000
50%32.00000038.95000040000.000000
75%38.00000039.17500052000.000000
max39.00000041.10000085000.000000

describe() lists out different descriptive statistical measures for all numerical columns in our dataset. By assigning the include attribute the value ‘all’, we can get the description to include all columns, including those containing categorical information.

4. memory_usage()

memory_usage() returns a Pandas Series having the memory usage of each column (in bytes) in a Pandas DataFrame. By specifying the deep attribute as True, we can get to know the actual space being taken by each column. More details on memory_usage() can be found here.

data_1.memory_usage(deep=True)

Output:

Index 80 Name 559 Age 72 City 578 State 584 DOB 603 Gender 553 City temp 72 Salary 72 dtype: int64

The memory usage of each column has been given as output in a Pandas Series. It is important to know the memory usage of a DataFrame, so that you can tackle errors like MemoryError in Python.

5. astype()

astype() is used to cast a Python object to a particular data type. It can be a very helpful function in case your data is not stored in the correct format (data type). For instance, if floating point numbers have somehow been misinterpreted by Python as strings, you can convert them back to floating point numbers with astype(). Or if you want to convert an object datatype to category, you can use astype().

data_1['Gender'] = data_1.Gender.astype('category')

You can verify the change in data type by looking at the data types of all columns in the dataset using the dtypes attribute. For looking at the documentation for astype(), click here.

6. loc[:]

loc[:] helps to access a group of rows and columns in a dataset, a slice of the dataset, as per our requirement. For instance, if we only want the last 2 rows and the first 3 columns of a dataset, we can access them with the help of loc[:]. We can also access rows and columns based on labels instead of row and column number.

data_1.loc[0:4, ['Name', 'Age', 'State']]

Output:

NameAgeState
0Alam29Madhya Pradesh
1Rohit23Delhi
2Bimla35Haryana
3Rahul25West Bengal
4Chaman32Tamil Nadu

The above code will return the “Name”, “Age”, and “State” columns for the first 5 customer records. Keep in mind that index starts from 0 in Python, and that loc[:] is inclusive on both values mentioned. So 0:4 will mean indices 0 to 4, both included.

loc[:] is one of the most powerful functions in Pandas, and is a must-know for all Data Analysts and Data Scientists. You can find the documentation for loc[:] here.

iloc[:] works in a similar manner, just that iloc[:] is not inclusive on both values. So iloc[0:4] would return rows with index 0, 1, 2, and 3, while loc[0:4] would return rows with index 0, 1, 2, 3, and 4. The documentation for iloc[:] can be found here.

7. to_datetime()

to_datetime() converts a Python object to datetime format. It can take an integer, floating point number, list, Pandas Series, or Pandas DataFrame as argument. to_datetime() is very powerful when the dataset has time series values or dates.

data_1['DOB'] = pd.to_datetime(data_1['DOB'])

The DOB column has now been changed to Pandas datatime format. All datetime functions can now be applied on this column. You can read more about to_datetime() here.

8. value_counts()

value_counts() returns a Pandas Series containing the counts of unique values. Consider a dataset that contains customer information about 5,000 customers of a company. value_counts() will help us in identifying the number of occurrences of each unique value in a Series. It can be applied to columns containing data like State, Industry of employment, or age of customers.

data_1['State'].value_counts()

Output:

Haryana 3 Delhi 2 West Bengal 1 Tamil Nadu 1 Bihar 1 Madhya Pradesh 1 Name: State, dtype: int64

The number of occurrences of each state in our dataset has been returned in the output, as expected. value_counts() can also be used to plot bar graphs of categorical and ordinal data.

data_1['State'].value_counts(normalize=True).plot(kind='bar', title='State')

The documentation for value_counts() can be found here.

9. drop_duplicates()

drop_duplicates() returns a Pandas DataFrame with duplicate rows removed. Even among duplicates, there is an option to keep the first occurrence (record) of the duplicate or the last. You can also specify the inplace and ignore_index attribute.

data_1.drop_duplicates(inplace=True)

inplace=True makes sure the changes are applied to the original dataset. You can verify the changes by looking at the shape of the original dataset, and the modified dataset (after dropping duplicates). You will notice the number of rows have reduced from 9 to 8 (because 1 duplicate has been dropped).

10. groupby()

groupby() is used to group a Pandas DataFrame by 1 or more columns, and perform some mathematical operation on it. groupby() can be used to summarize data in a simple manner.

data_1.groupby(by='State').Salary.mean()

Output:

State Bihar 18000 Delhi 68500 Haryana 27500 Madhya Pradesh 50000 Tamil Nadu 65000 West Bengal 40000 Name: Salary, dtype: int64

The above code will group the dataset by “State” column, and will return the mean age across states. You can click here to know more about groupby().

11. merge()

merge() is used to merge 2 Pandas DataFrame objects or a DataFrame and a Series object on a common column (field). If you are familiar with the concept of JOIN in SQL, merge function similar to that. It returns the merged DataFrame.

data_1.merge(data_2, on='Name', how='left')

To know more about attributes like on (including left_on and right_on), how, and suffixes, refer to the documentation.

12. sort_values()

sort_values() is used to sort column in a Pandas DataFrame (or a Pandas Series) by values in ascending or descending order. By specifying the inplace attribute as True, you can make a change directly in the original DataFrame.

data_1.sort_values(by='Name', inplace=True)

Output:

NameAgeCityStateDOBGenderCity tempSalary
0Alam29IndoreMadhya Pradesh1991-11-20Male35.550000
2Bimla35RohtakHaryana1985-09-01Female39.720000
4Chaman32ChennaiTamil Nadu1988-12-03Male41.165000
6Charu29New DelhiDelhi1992-03-18Female39.052000
7Ganesh39PatnaBihar1981-07-12MaleNaN18000
3Rahul25KolkataWest Bengal1995-09-19Male36.540000
1Rohit23New DelhiDelhi1997-09-19Male39.085000
5Vivek38GurugramHaryana1982-06-22Male38.935000

You can see that the ordering of records has changed now. Records are now listed in alphabetical order of Names. sort_values() has many other attributes which can be specified. You can read about it here.

Similar to sort_values() is sort_index(). It is used to sort the DataFrame by index instead of a column value.

13. fillna()

Typically in a large dataset, you will find several entries labelled NaN by Python. NaN stands for “not a number”, and represents entries that were not populated in the original data source. While populating the values in the DataFrame, Pandas makes sure that these entries can be identified separately by the user.

fillna() helps to replace all NaN values in a DataFrame or Series by imputing these missing values with more appropriate values.

data_1['City temp'].fillna(38.5, inplace=True)

The above code will replace all blank “City temp” entries with 38.5. The missing values could be imputed with the mean, median, mode, or some other value. We have chosen mean for our case.

You may also like

Leave a Reply

[script_16]