This set of documentation demonstrates how to connect to WRDS server for data extraction. If your institution is a WRDS subscriber, you can log in using your usual username and password once prompted in the Python code.
0. Setting Up
Preparation - Install WRDS Python Module
“All WRDS data is stored in a PostgreSQL database, and is available through Python through our in-house Python module, wrds, which is freely available on PyPI via a pip install.”
https://pypi.org/project/wrds/
Once you have the module installed, you can then use any of the common Python programming interface (Spyder, Jupyter Notebook, JupyterLab, etc).
1. Connecting to WRDS
1.1 Establish Connection
1.2 Functions - List Libraries
Depending on your institution's data subscription, you will see a list of all data libraries available to you.
1.3 Functions - List Tables Within a Library
This command will list all the datasets that is available under the CRSP database, which researchers often refer to for pricing related data.
1.4 Explore a Given Table
This command returns a summary table of the "stocknames" dataset under the "crsp" database. This is similar to "proc content" command in SAS. If you are not already familiar with what is inside a particular dataset, this command is a good way to get a quick overview.
2. Calling Data
There are various ways to extract data through the WRDS API. The basic method is good for extracting data from one single dataset, while the sql based method is more flexible in joining multiple data sources and imposing conditioning statements.
2.1 Basic Method
The command lines extract first 1000 observations from the table "stocknames" under "crsp" library.
This is equivalent to the SAS lines:
2.2 Advanced Method
raw_sql function gives users more flexibility in extracting, joining, filtering data.
Example 1: calling data while imposing conditioning statement:
Example 2: calling data from more than one data source while imposing conditional statement
If you are already an experienced "proc sql" user under SAS, then these advanced data extraction lines should look fairly intuitive to you. Just keep in mind the subtle difference in the date formats accepted here.
3. Storing Output
Last but not least, when the program is done running and outputs need to be stored, there are several options.
3.1 Native "Pickle"
The Python native "pickle" package is versatile for storing and retrieving data. The code below stores the data "crsp_m" extracted from the raw_sql method above into a file "aapl.pkl".
When the data is needed in other programs, one can easily retrieve it by using the following command lines. The content of the data is now loaded as data frame "aapl_msf".
3.2 Flat Files
If you would like to port the output to other programming languages for further data crunching, you can store it using the lines below:
There are various other packages that can enable data storage in other formats, for instance, SASPy.