showml 1.4.1

A Python package of Machine Learning Algorithms implemented from scratch

ShowML

Show the ML Code!

A Python package of Machine Learning Algorithms implemented from scratch.

The aim of this package is to present the working behind fundamental Machine Learning algorithms in a transparent and modular way.

NOTE: The implementations of these algorithms are not thoroughly optimized for high computational efficiency.

Usage

Check out: showml/examples/.

Installation

Install the package:

$ pip install showml

To clone the repository and view the source files:

$ git clone https://github.com/hasnainroopawalla/ShowML.git
$ cd ShowML
$ pip install -r requirements.txt

How to Contribute

Contents

Models

Linear

• Linear Regression (showml.linear_model.regression.LinearRegression)
• Logistic Regression (showml.linear_model.regression.LogisticRegression)

Non-Linear

• Sequential (showml.deep_learning.model.Sequential)

Deep Learning

Layers

• Dense (showml.deep_learning.layers.Dense)

Activations

• Sigmoid (showml.deep_learning.activations.Sigmoid)
• ReLu (showml.deep_learning.activations.Relu)
• Softmax (showml.deep_learning.activations.Softmax)

Optimizers

• Stochastic/Batch/Mini-Batch Gradient Descent (showml.optimizers.SGD)
• Adaptive Gradient (showml.optimizers.AdaGrad)
• Root Mean Squared Propagation (showml.optimizers.RMSProp)

Loss Functions

• Mean Squared Error (showml.losses.MeanSquaredError)
• Binary Cross Entropy (showml.losses.BinaryCrossEntropy)
• Categorical Cross Entropy (showml.losses.CrossEntropy)

Simulations

• Conway's Game of Life (showml.simulations.conways_game_of_life.GameOfLife)

Contributing

1. Fork the repository.
2. Install the necessary dependencies:

$ pip install pre-commit mypy pytest

3. Commit and push your changes to your own branch.
4. Before submitting a Pull Request, run these housekeeping checks locally:-

• Run pre-commit:

$ pre-commit run -a

• Run mypy:

$ mypy .

• Run pytest:

$ pytest

5. Once everything succeeds, create a Pull Request (CI will be triggered)