marcodavidg / mlpackage

About the project

Part of the MLOps Bootcamp: Mastering AI Operations for Success - AIOps Udemy course.

https://www.udemy.com/course/mlops-bootcamp-mastering-ai-operations-for-success-aiops

Technologies used

• Python
• Sci-kit learn
• Pytest
• Pandas
• setuptools

Packaging the ML Model of Classification

Problem Statement

• Company wants to automate the loan eligibility process based on customer detail provided while filling online application form.
• It is a classification problem where we have to predict whether a loan would be approved or not.

Data

The data corresponds to a set of financial requests associated with individuals.

Variables	Description
Loan_ID	Unique Loan ID
Gender	Male/ Female
Married	Applicant married (Y/N)
Dependents	Number of dependents
Education	Applicant Education (Graduate/ Under Graduate)
Self_Employed	Self employed (Y/N)
ApplicantIncome	Applicant income
CoapplicantIncome	Coapplicant income
LoanAmount	Loan amount in thousands
Loan_Amount_Term	Term of loan in months
Credit_History	credit history meets guidelines
Property_Area	Urban/ Semi Urban/ Rural
Loan_Status	Loan approved (Y/N)

Source: Kaggle

Directory structure

prediction_model


├── MANIFEST.in
├── prediction_model
│   ├── config
│   │   ├── config.py
│   │   └── __init__.py
│   ├── datasets
│   │   ├── __init__.py
│   │   ├── test.csv
│   │   └── train.csv
│   ├── __init__.py
│   ├── pipeline.py
│   ├── predict.py
│   ├── processing
│   │   ├── data_handling.py
│   │   ├── __init__.py
│   │   └── preprocessing.py
│   ├── trained_models
│   │   ├── classification.pkl
│   │   └── __init__.py
│   ├── training_pipeline.py
│   └── VERSION
├── README.md
├── requirements.txt
├── setup.py
└── tests
    ├── pytest.ini
    └── test_prediction.py

Build the Package

1. Goto Project directory and install dependencies pip install -r requirements.txt

2. Create Pickle file after training: python prediction_model/training_pipeline.py

3. Create source distribution and wheel python setup.py sdist bdist_wheel

Installation of Package

Go to project directory where setup.py file is located

1. To install it in editable or developer mode

pip install -e .

. refers to current directory

-e refers to --editable mode

2. Normal installation

pip install .

. refers to current directory

3. Also can be installed from git as well after pushing to github

pip install git+https://github.com/marcodavidg/MLPackage.git

Testing the Package Working

1. Remove the PYTHONPATH from environment variables
2. Go to a separate location which is outside of package directory
3. Create a new virual environment & activate it
4. Before installing, test whether you are able to import the package of prediction_model - (you should not be able to do it)
5. Now in the new environment install the package from github pip install git+https://github.com/marcodavidg/MLPackage.git
6. Now try importing the prediction_model, you should be able to do it successfully
7. Extras : Run training pipeline using the package, and also conduct the test