skleam-code 0.3.0a0

A plug-and-play machine learning package with and without libraries.

skleam-code

A comprehensive, plug-and-play machine learning library offering algorithm implementations both with and without standard libraries (from scratch). It is designed to be highly educational, easy to use, and immediately functional.

Installation

Install the package via pip:

pip install skleam-code

Overview

The skleam-code library is structured into two main sub-modules:

• with_lib: Models built on top of standard libraries like scikit-learn.
• no_lib: Models built strictly from scratch using pure numpy mathematics.

Both categories contain the exact same lineup of 13 algorithms!

Supported Algorithms:

• Regression: Linear Regression, Logistic Regression
• Tree-Based: Decision Tree, Random Forest, Bagging Random Forest
• Boosting: AdaBoost, XGBoost
• SVM: Linear SVM, Non-Linear SVM
• Unsupervised / Dimensionality Reduction: K-Means Clustering, PCA, SVD, LDA

---

Quick Start & Usage

Every single model in skleam-code shares a uniform, plug-and-play API. You only need to know two methods to use any algorithm:

• .fit(X, y) or .fit(X) for unsupervised models.
• .predict(X) or .transform(X) for projection models.

Example: Using Pre-built "From Scratch" Models (no_lib)

from skleam.no_lib import DecisionTreeNoLib

# 1. Initialize the model
model = DecisionTreeNoLib(max_depth=5)

# 2. Automatically generate dummy data to test it out!
X, y = model.generate_data(n_samples=200, n_features=4, task='classification')

# 3. Fit and Predict
model.fit(X, y)
predictions = model.predict(X)
print("Predictions:", predictions[:10])

Example: Using Standard Library Backed Models (with_lib)

from skleam.with_lib import RandomForestWithLib

# Initialize the model (you can pass kwargs supported by the underlying sklearn model)
rf = RandomForestWithLib(n_estimators=100, max_depth=3)

# Generate dummy data
X, y = rf.generate_data(task='classification')

# Fit and Predict
rf.fit(X, y)
print("RF Predictions:", rf.predict(X)[:5])

---

Generating Artificial Test Data

One of the best features of skleam-code is the built-in static method generate_data(), attached to every single model. If you don't have a dataset ready, the library will generate one specifically formatted for the task at hand.

Supported tasks include: 'classification', 'regression', 'clustering', and 'unsupervised'.

from skleam.no_lib import KMeansNoLib

# Generate clustering data instantly
X, y = KMeansNoLib.generate_data(n_samples=300, n_features=2, task='clustering')

kmeans = KMeansNoLib(K=3)
kmeans.fit(X)
print(kmeans.predict(X)[:10])

Exploring the Architecture

You can quickly view all available algorithms directly in your terminal using the built-in index tool:

from skleam import index
index.main()