MLflow Classes for Credit Risk Prediction and Image Classification

This repository contains a set of classes for training machine learning models and logging experiments using MLflow library. The classes are designed to work with structured data for credit risk prediction and image classification problems.

Usage

Credit Risk Prediction

The repository contains a Jupyter notebook main.ipynb that demonstrates how to use the classes to train a credit risk prediction model using a structured dataset. The notebook shows how to:

• Load data and perform preprocessing
• Train and evaluate a machine learning model
• Log the parameters, metrics, and artifacts using MLflow

Image Classification

The repository also contains a Jupyter notebook image_classification.ipynb that demonstrates how to use the classes to train an image classification model using a pre-trained EfficientNet model. The notebook shows how to:

• Load and preprocess images
• Fine-tune a pre-trained model
• Log the parameters, metrics, and artifacts using MLflow

Classes

The repository contains the following classes:

• CreditRiskExperiment: A class for running a credit risk prediction experiment using a structured dataset.
• ImageClassificationExperiment: A class for running an image classification experiment using a pre-trained EfficientNet model.

Both classes have the following methods:

• __init__: Initializes the class and sets up the MLflow experiment.
• run: Runs the experiment and logs the parameters, metrics, and artifacts.
• log_params: Logs the parameters used in the experiment.
• log_metrics: Logs the metrics computed during the experiment.
• log_artifacts: Logs the artifacts generated during the experiment.

Installation

To use the classes, first clone the repository:

git clone https://github.com/<username>/<repository-name>.git