RubyScientistAndGraphics

Lightweight data science toolkit for Ruby: load/clean data, get quick stats, plot charts, and train simple ML models — all in one gem and with zero heavy dependencies.

It ships with a minimal in-house DataFrame (no Daru required), Gruff for plotting, and tiny implementations for statistics and ML (linear regression and k-means).

• Load and save CSV/JSON, plus save/load a simple “project” (columns + rows).
• Data cleaning helpers: remove columns, fill missing values, limit rows.
• Quick stats: per-column mean/min/max and Pearson correlation.
• Plotting: bar and line charts via Gruff.
• ML: linear regression (least squares) and k-means clustering.

Clone and use directly, or add to your Gemfile from a git source until published to RubyGems:

gem 'ruby_scientist_and_graphics', git: 'https://github.com/your-user/ruby_scientist_and_graphics'

Then install:

bundle install

Ruby 3.2+ is recommended.

Run the demo to see the workflow end-to-end:

ruby demo.rb

Or use the API:

require_relative 'lib/ruby_scientist_and_graphics'

interface = RubyScientistAndGraphics::Interface.new

# 1) Load and clean
interface.load('test/fixtures/sample.csv', remove_columns: [:comentarios], limit: 5)
interface.clean(missing: 0)

# 2) Stats
interface.analyze

# 3) Plot
interface.graph(type: :bar, x: :mes, y: :ventas, file: 'output.png')

# 4) Train a model
model = interface.train_model(type: :linear_regression, features: [:mes], target: :ventas)
preds = model.predict([[1.0], [2.0], [3.0]])

# 5) Save project
interface.save_project('project.json')

# 6) Load a previously saved project and predict
interface.load_project('project.json')
interface.train_model(type: :linear_regression, features: [:mes], target: :ventas)
preds = interface.predict([[6.0], [7.0]])

• DataFrame (internal): CSV load, indexing by column symbol, head, write_csv, map_vectors, filter_rows.
• IO: load_csv, load_json, save_csv, save_json, save_project, load_project.
• Dataset: remove_columns, add_column, limit_rows, fill_missing, head, stats, plot.
• Stats: describe, correlation(col1, col2).
• Plotter: bar(x:, y:, file:), line(x:, y:, file:).
• Interface: load, clean, analyze, graph, pipeline, train_model, save_project.
• Interface: load, clean, analyze, graph, pipeline, train_model, save_project, load_project, predict.

This gem includes a minimal in-house DataFrame that powers all features. If you want more performance or richer operations (group-by, joins, rolling, etc.), you can plug a third-party backend behind the same API using a simple adapter pattern.

Potential backends:

• Polars (Ruby bindings): very fast, columnar engine written in Rust.
• Rover-Df: pure Ruby DataFrame with a friendly API.

Adapter idea (sketch):

module RubyScientistAndGraphics
	module Backends
		class PolarsAdapter
			def self.from_csv(path); end
			def vectors; end
			def [](col); end
			def to_a; end
			# implement methods used by Dataset/Stats/Plotter
		end
	end
end

# Then inject at app start:
# RubyScientistAndGraphics::DataFrame = RubyScientistAndGraphics::Backends::PolarsAdapter

This keeps your app code unchanged while letting you switch engines.

Setup and tests:

bin/setup
bundle exec rake test

Run an interactive console:

bin/console

Build and install locally:

bundle exec rake install

Release flow: bump version in lib/ruby_scientist_and_graphics/version.rb, then:

bundle exec rake release

Pull requests are welcome. Please open an issue to discuss large changes first. See CODE_OF_CONDUCT.md.

MIT License. See LICENSE.txt.