mlguide 1.0.0

Modular ML toolkit with a built-in guide system for students.

mlguide 🚀

The ML toolkit that teaches you while it works.

mlguide is a modular, student-first ML toolkit with three layers:

1. Guided Learning — a built-in interactive help system that explains ML concepts.
2. Individual Modules — every pipeline stage is independently importable.
3. Full Autopilot — run_pipeline() chains everything together.

Installation

pip install mlguide

Quick Start

Autopilot (one line)

from mlguide import sample_data, run_pipeline

df = sample_data("regression")
result = run_pipeline(df, target="price")

Step by Step

from mlguide import load_data, clean, split, encode, scale, train, evaluate

df = load_data("housing.csv")
df = clean(df, target="price")
X_tr, X_te, y_tr, y_te = split(df, target="price")
X_tr, enc = encode(X_tr, fit=True)
X_te, _   = encode(X_te, encoder=enc)
X_tr, sc  = scale(X_tr, fit=True)
X_te, _   = scale(X_te, scaler=sc)
model     = train(X_tr, y_tr, model="random_forest")
metrics   = evaluate(model, X_te, y_te)

Need Help?

from mlguide import guide

guide()              # overview
guide("ml_basics")   # what is ML?
guide("split")       # why we split data
guide("train")       # all available models
guide("cheatsheet")  # compact reference

Features

• Zero data leakage — split before encode/scale is enforced by architecture.
• Transparent — every function logs what it did and why.
• Modular — use any function independently.
• Student-friendly errors — typo suggestions, available column listings, clear next-step guidance.
• Lightweight — only pandas, numpy, scikit-learn, and joblib.
• Bundled datasets — sample_data("regression") for instant practice.
• Text & NLP Extraction — builtin regex extraction and preprocessing without external NLP libraries.

API Reference

Function	Description
guide(topic)	Interactive help system
run_pipeline(source, target)	Full autopilot pipeline
load_data(source)	Load CSV or DataFrame
sample_data(name)	Bundled practice datasets
clean(df, target)	Clean data (nulls, duplicates, cardinality)
split(df, target)	Train/test split with auto-stratification
encode(X, fit=True)	One-Hot Encode categorical columns
scale(X, method)	Scale numeric features
detect_task(y)	Infer regression vs classification
train(X, y, model)	Train a model
compare_models(X, y)	Cross-validated model comparison
evaluate(model, X, y)	Evaluate on test set
predict(model, data)	Make predictions
save_model(model, path)	Save model bundle
load_model(path)	Load model bundle
get_feature_importance(model)	Feature importance table
extract_emails(), extract_phones()	Regex-based extraction
clean_text()	Full NLP preprocessing pipeline

License

MIT — Joel Inian Francis