Environment management with conda and pip in MLOps - Mini Project: Build & Apply

Environment management with conda and pip

📖 Scenario: You are working on a machine learning project that requires specific Python packages. To keep your project organized and avoid conflicts with other projects, you will create and manage a separate environment using conda and pip.

🎯 Goal: Learn how to create a new conda environment, install packages using pip, and verify the installed packages.

📋 What You'll Learn

Create a new conda environment named ml_project_env with Python 3.12

Install the numpy package using conda

Install the scikit-learn package using pip inside the conda environment

List all installed packages in the environment to verify

💡 Why This Matters

🌍 Real World

Managing Python environments helps avoid conflicts between projects and keeps your machine learning projects organized.

💼 Career

Environment management is a key skill for data scientists and MLOps engineers to ensure reproducible and stable workflows.

1

Create a new conda environment

Type the command to create a new conda environment named ml_project_env with Python version 3.12.

MLOps

# Your code here

Hint

Use conda create -n <env_name> python=<version> to create the environment.

2

Activate the conda environment and install numpy

Type the command to activate the conda environment named ml_project_env. Then type the command to install the numpy package using conda.

MLOps

conda create -n ml_project_env python=3.12 -y
# Your code here

Hint

Use conda activate ml_project_env to activate. Use conda install numpy to install numpy.

3

Install scikit-learn using pip inside the environment

Type the command to install the scikit-learn package using pip inside the activated ml_project_env environment.

MLOps

conda activate ml_project_env
conda install numpy -y
# Your code here

Hint

Use pip install scikit-learn to install the package inside the environment.

4

List installed packages to verify

Type the command to list all installed packages in the ml_project_env environment using conda list.

MLOps

pip install scikit-learn
# Your code here

Hint

Use conda list to see all installed packages in the environment.

Practice

1. What is the main purpose of using conda create -n myenv python=3.8?

easy

A. To delete the environment named 'myenv' and install Python 3.8 globally

B. To update Python to version 3.8 in the current environment

C. To create a new isolated environment named 'myenv' with Python 3.8 installed

D. To install all packages listed in a file named 'myenv' with Python 3.8

Solution

1. Step 1: Understand the conda create command

   This command is used to create a new environment in conda, isolating packages and Python versions.

2. Step 2: Analyze the flags and arguments

   The -n myenv specifies the environment name, and python=3.8 sets the Python version inside it.

3. Final Answer:

   To create a new isolated environment named 'myenv' with Python 3.8 installed -> Option C

4. Quick Check:

   conda create -n myenv python=3.8 = D [OK]

Hint: Remember: 'conda create -n' makes new isolated envs [OK]

Common Mistakes:

• Confusing 'create' with 'install' or 'update'
• Thinking it affects the global Python installation
• Misunderstanding the '-n' flag as package name

2. Which of the following commands correctly activates a conda environment named dataenv?

easy

A. activate conda dataenv

B. conda activate dataenv

C. conda start dataenv

D. source deactivate dataenv

Solution

1. Step 1: Recall the syntax to activate conda environments

   The correct command to activate an environment is conda activate <env_name>.

2. Step 2: Check each option

   conda activate dataenv matches the correct syntax. Options B, C, and D use incorrect command order or wrong commands.

3. Final Answer:

   conda activate dataenv -> Option B

4. Quick Check:

   Activate env = conda activate env_name [OK]

Hint: Use 'conda activate env_name' to switch environments [OK]

Common Mistakes:

• Using 'activate conda' instead of 'conda activate'
• Confusing 'source deactivate' with activation
• Trying 'conda start' which is invalid

3. Given the following commands run in order:

conda create -n testenv python=3.9 -y
conda activate testenv
pip install numpy
pip list | grep numpy

What will be the output of the last command?

medium

A. numpy with its installed version number

B. No output because pip list does not work inside conda

C. Error: 'pip' command not found

D. List of all packages except numpy

Solution

1. Step 1: Understand environment creation and activation

   The environment 'testenv' is created with Python 3.9 and then activated, so all commands run inside it.

2. Step 2: Installing numpy with pip inside the active environment

   Running pip install numpy installs numpy in 'testenv'. The pip list | grep numpy command will show numpy and its version.

3. Final Answer:

   numpy with its installed version number -> Option A

4. Quick Check:

   pip install inside active env = numpy listed [OK]

Hint: pip installs packages in active conda env, visible with pip list [OK]

Common Mistakes:

• Thinking pip installs globally ignoring conda env
• Assuming pip commands fail inside conda
• Expecting no output from pip list

4. You run conda activate myenv but get the error: CommandNotFoundError: Your shell has not been properly configured to use 'conda activate'. What is the most likely fix?

medium

A. Reinstall Python globally on your system

B. Deactivate any active environment before activating 'myenv'

C. Use pip install conda to fix the error

D. Run conda init to configure your shell, then restart the terminal

Solution

1. Step 1: Understand the error cause

   This error means the shell does not know how to run conda activate because it lacks proper initialization.

2. Step 2: Apply the fix by initializing conda for the shell

   Running conda init sets up the shell scripts needed. Restarting the terminal applies changes.

3. Final Answer:

   Run conda init to configure your shell, then restart the terminal -> Option D

4. Quick Check:

   Shell config for conda = conda init + restart [OK]

Hint: Run 'conda init' once after install, then restart terminal [OK]

Common Mistakes:

• Trying to reinstall Python instead of fixing shell config
• Using pip to install conda which is incorrect
• Ignoring the need to restart terminal after init

5. You want to create a reproducible environment for a project using conda and pip. Which sequence of commands correctly creates an environment, installs packages from a requirements.txt file using pip, and exports the environment including pip packages?

hard

A. conda create -n projenv python=3.10 -y && conda activate projenv && pip install -r requirements.txt && conda env export > environment.yml

B. conda create -n projenv python=3.10 -y && pip install -r requirements.txt && conda activate projenv && conda env export > environment.yml

C. conda create -n projenv python=3.10 -y && conda activate projenv && pip install -r requirements.txt && conda env export --from-history > environment.yml

D. conda activate projenv && conda create -n projenv python=3.10 -y && pip install -r requirements.txt && conda env export > environment.yml

Solution

1. Step 1: Create and activate the environment before installing packages

   You must first create the environment, then activate it to install packages inside it.

2. Step 2: Install pip packages and export full environment

   After activation, install packages from requirements.txt using pip. Then export the full environment including pip packages with conda env export.

3. Final Answer:

   conda create -n projenv python=3.10 -y && conda activate projenv && pip install -r requirements.txt && conda env export > environment.yml -> Option A

4. Quick Check:

   Create, activate, pip install, export full env = C [OK]

Hint: Activate env before pip install; export full env to include pip packages [OK]

Common Mistakes:

• Installing pip packages before activating environment
• Using --from-history which excludes pip packages
• Activating environment after installing packages