One of the main obstacles of Data Engineering is the large and varied technical skills that can be required on a day-to-day basis.
*** Note - If you email a link to your GitHub repo with all the completed exercises, I will send you back a free copy of my ebook Introduction to Data Engineering. ***
This aim of this repository is to help you develop and learn those skills. Generally, here are the high level topics that these practice problems will cover.
- Python data processing.
- csv, flat-file, parquet, json, etc.
- SQL database table design.
- Python Postgres, data ingestion and retrieval.
- PySpark
- Data cleansing / dirty data.
You will need two things to work effectively on most all of these problems.
Dockerdocker-compose
All the tools and technologies you need will be packaged
into the dockerfile for each exercise.
For each exercise you will need to cd into that folder and
run the docker build command, that command will be listed in
the README for each exercise, follow those instructions.
The first exercise tests your ability to download a number of files
from an HTTP source and unzip them, storing them locally with Python.
cd Exercises/Exercise-1 and see README in that location for instructions.
The second exercise
tests your ability perform web scraping, build uris, download files, and use Pandas to
do some simple cumulative actions.
cd Exercises/Exercise-2 and see README in that location for instructions.
The third exercise tests a few skills.
This time we will be using a popular aws package called boto3 to try to perform a multi-step
actions to download some open source s3 data files.
cd Exercises/Exercise-3 and see README in that location for instructions.
The fourth exercise
focuses more file types json and csv, and working with them in Python.
You will have to traverse a ragged directory structure, finding any json files
and converting them to csv.
The fifth exercise
is going to be a little different than the rest. In this problem you will be given a number of
csv files. You must create a data model / schema to hold these data sets, including indexes,
then create all the tables inside Postgres by connecting to the database with Python.
The sixth exercise
Is going to step it up a little and move onto more popular tools. In this exercise we are going
to load some files using PySpark and then be asked to do some basic aggregation.
Best of luck!
The seventh exercise
Taking a page out of the previous exercise, this one is focus on using a few of the
more common build in PySpark functions pyspark.sql.functions and applying their
usage to real-life problems.
Many times to solve simple problems we have to find and use multiple functions available from libraries. This will test your ability to do that.
The eighth exercise Using new tools is imperative to growing as a Data Engineer. DuckDB is one of those new tools. In this exercise you will have to complete a number of analytical and transformation tasks using DuckDB. This will require an understanding of the functions and documenation of DuckDB.
The ninth exercise Polars is a new Rust based tool with a wonderful Python package that has taken Data Engineering by storm. It's better than Pandas because it has both SQL Context and supports Lazy evalutation for larger than memory data sets! Show your Lazy skills!
The tenth exercise This exercise is to help you learn Data Quality, specifically a tool called Great Expectations. You will be given an existing datasets in CSV format, as well as an existing pipeline. There is a data quality issue and you will be asked to implement some Data Quality checks to catch some of these issues.