Welcome to ML In Production 2020 edition’s documentation!

Table of Contents

• 1. Environment management
  • 1.1. Why should we care?
    • 1.1.1. How we typically think of our work
    • 1.1.2. Even the simplest code relies on many other things…
    • 1.1.3. And needs to run on different environments
    • 1.1.4. How we should think of our work
  • 1.2. Python dependency management
    • 1.2.1. Many snakes…
    • 1.2.2. How does Python imports libraries?
    • 1.2.3. What does pip install do?
    • 1.2.4. Example: numpy
    • 1.2.5. Pip hell
    • 1.2.6. We need a tool to:
    • 1.2.7. Available tools
  • 1.3. Env Management via Conda
    • 1.3.1. What is conda?
    • 1.3.2. Quick foreplay
    • 1.3.3. You can port conda environments
    • 1.3.4. You can manage non-python dependencies
    • 1.3.5. You can continue using pip… with care
    • 1.3.6. conda envs limitations
    • 1.3.7. Quizz
    • 1.3.8. Quizz (solution)
    • 1.3.9. Practice time!
    • 1.3.10. Practice time (solution)
    • 1.3.11. Additional resources
• 2. Simple automation with Make
  • 2.1. Makefiles
    • 2.1.1. Make “Hello World!”
    • 2.1.2. Make concepts
    • 2.1.3. Using variables
    • 2.1.4. Dry-run: seeing the execution plan
    • 2.1.5. Parallel execution
    • 2.1.6. Make gotchas
    • 2.1.7. Using Make with conda
    • 2.1.8. Quizz I
    • 2.1.9. Quizz I: solution
    • 2.1.10. Quizz II
    • 2.1.11. Quizz II: solution
    • 2.1.12. Practice time
    • 2.1.13. Practice time (solution)
• 3. Project templates
  • 3.1. Templates
    • 3.1.1. Most DS projects consist in similar workflows
    • 3.1.2. Example project structure
    • 3.1.3. cookiecutter Hello world!
    • 3.1.4. Recommendations
    • 3.1.5. cookiecutter: pre/post hooks
    • 3.1.6. cookiecutter: Jinja Templates
    • 3.1.7. Practice time: Build your own template!
• 4. Programming in Python for DS
  • 4.1. OOP in Python
    • 4.1.1. Classes and objects
    • 4.1.2. Classes have members
    • 4.1.3. Defining class and instance attributes
    • 4.1.4. Defining instance methods
    • 4.1.5. Defining class methods
    • 4.1.6. Inheritance
    • 4.1.7. Multiple Inheritance
    • 4.1.8. Mixin classes
    • 4.1.9. Abstract classes and interfaces
    • 4.1.10. Why bother about all of this!?
    • 4.1.11. Python’s philosophy: (Dynamic) Duck typing
    • 4.1.12. Practice time
  • 4.2. Packaging in Python
    • 4.2.1. Modules
    • 4.2.2. Packages
    • 4.2.3. setuptools
    • 4.2.4. Practice time
  • 4.3. Other useful Python patterns
    • 4.3.1. decorators
    • 4.3.2. iterators
    • 4.3.3. generators