anaconda-notes

Notes on Anaconda

anaconda cloud

freelearning.anaconda.cloud

What are Modules, Packages, and Libraries?:

@01:13: Tools

Data Analysis

• pandas
• NumPy
• SciPy

Machine Learning

• TensorFlow
• scikit learn
• PyTorch
• Keras

Data Visualization

• matplotlib
• bokeh
• plotly
• Streamlit

Understanding Conda:

conda: an open-source tool for managing:

• package:
  • repositories
  • dependencies and
• environments for all programming languages:
• Python
• R
• Ruby
• Lua
• Scala
• Java
• JavaScript
• C/C++
• FORTRAN

@00:21:

A cooking analogy:

• conda like a kitchen

• has pantry - store utensils, appliances, ingredients

• in conda - package repository acts like a pantry

• where instead of utensils, appliances, ingredients you find softwaree programs known as packages.

• a dish depends on utensils and ingredients

• in conda each package may depend on other packages

• e.g. when install package: pandas, automatically install other packages that depends on like NumPy.

@01:13:

• packages are like recipies
• you may use one prepared by someone else with recipe and ingredients. -> reproducible meal
• in Anaconda environments act like meal kits
• create a reproducible environment from an environment.yml file you or others created
• can pin versions of packages in environment.

@01:51: Recommended: Create separate environments per project

Environment management:

• environment each for:
  • ML/AI Project
  • ETL dashboarding project
  • R project

Conda Workflow: Creating Environments, Installing Packages, and Launching an IDE

@00:18:

A Conda Workflow:

1. create environment
2. activate environment
3. install packages
4. launch JupyterLab
5. deactivate environment

@00:34:

1. Create environment

on Windows: open anaconda prompt

conda --version

conda env list

• you always begin in base environment
• always create a new environment
• never work in base

@01:07

conda create --name example

can create environment with a specific version of python:

e.g. :

conda create --name examnple python=3.9

@01:41

2. Activate Environment

moves you out of base into -> new environment e.g. example

• here you can install separate
• distinct versions of python
• and packages
• specifically for your project
• without affecting base environment packages.

Rerun:

conda env list

ensure new environment: example is in your environment list

@02:04:

to activate:

conda activate example

@02:12:

once in new environment:

conda list

3. install packages

to see installed packages

can install multiple packages in one command:

e.g. :

conda install jupyterlab dask pandas hvplot

@02:35:

If you cannot find a package on the default conda channel you can search for it on: anaconda.org where robust channels like conda forge host many additional packages

@02:45

to install a package from

conda forge

for example

add

-c

to the channel name

e.g. :

conda install -c conda-forge conda-stats

4. launch jupyterlab

jupyter-lab

@03:20 to @03:30

working in jupyter-lab in browser example

@03:30:

to end jupyter lab session:

Ctrl + C

@03:44:

5. deactivate environment

Best practice: Deactivate environment before switching to another environment

• once working with multiple projects

@03:52:

conda deactivate

What is an IDE? :

• IDE Integrated Development Environment
• Python IDE's

@00:24 :

An IDE has:

1. text editor
2. integrated build tools
   • for compiling languages like:
   • C, C++, Go, Rust
3. integrated debugger
   • tool helps fic errors in code

@00:39:

An IDE brings these 3 elements together in a single interface

@00:47:

Popular python IDE's

@01:19:

• Spyder has similarities with MATLAB

@01:27:

Jupyter:

• jupyter notebooks -- file browser -- notebook editor
• JupyterLab -- editing python and notebooks -- in a more IDE like environment

---

What is Jupyter Notebook and JupyterLab?