iragca 0.4.2

My custom scripts

iragca

My library of custom python scripts and configurations

---

---

iragca-python is a comprehensive Python library providing practical utilities for data science, machine learning, and visualization workflows. It streamlines common tasks in machine learning, data visualization, and functional programming.

Key Features

• Accessible Visualization: Professional matplotlib styles and WCAG-compliant color palettes designed for clarity and accessibility.
• Lightweight Experiment Tracking: RunLogger for logging metrics with dynamic property access and optional progress bars.
• Functional Programming Utilities: Composable data transformation pipelines using Pipeline and Step.
• Deprecation Management: Tools to manage deprecations and guide users to alternatives.

Use Cases

• ML/DL Training: Track metrics with RunLogger during training loops
• Data Pipelines: Build readable transformation chains with Pipeline
• Publication Plots: Create accessible visualizations with pre-configured styles
• Library Maintenance: Manage deprecations gracefully with proper warnings

Installation

Install using pip:

pip install iragca

Install a specific module (see the docs for options):

pip install iragca[functional]

Quick Start

RunLogger

from iragca.ml import RunLogger

logger = RunLogger(max_steps=100, display_progress=True)
for epoch in range(100):
	loss = 1.0 / (epoch + 1)
	logger.log_metrics({'loss': loss}, step=epoch)

print(f"Final loss: {logger.loss[-1]}")

Matplotlib Colors and Styles

import matplotlib.pyplot as plt

from iragca.matplotlib import Color, Styles

plt.style.use(Styles.CMR10.value)

sample_data = [1, 3, 2, 4, 3, 5]

plt.plot(sample_data, color=Color.BLUE.value)
plt.title("Sample Plot with Custom Style and Color")
plt.xlabel("X-axis")
plt.ylabel("Y-axis")
plt.show()

Functional Pipelines

from iragca.functional import Pipeline, Step

pipeline = Pipeline([
	lambda x: x * 2,
	Step(lambda x, n: x + n, n=10),
	lambda x: x ** 2,
])

result = pipeline(5)  # (5 * 2 + 10)^2 = 400

Documentation

Read the full documentation in the docs or visit the API reference.