kiwiml 0.0.4

a tiny, simple, instructional machine learning library in python

kiwi v0.0.4

A tiny, simple, instructional machine learning library in python, written by Ori Yonay.

To install: pip3 install kiwiml

Current Features:

• autodiff: A simple automatic numpy-compatible multidimensional differentiator
• Machine Learning Models:
  • KNN
  • Linear Regression
  • Logistic Regression
  • Naive Bayes
  • Perceptron
  • Single Dimensional Analysis
• Utilities:
  • Accuracy score function for calculating model accuracy
  • train_test_split
  • A function to plot cost histories
  • PCA Decomposition (will be in its own dimensionality reduction class in the future)
• Error Functions:
  • Mean-Squared Error
  • Cross-Entropy Loss

TODO:

• Error Functions: add mean absolute deviation
• Machine Learning Models:
  • SVM
  • Decision Tree
  • Random Forest
  • K-Means Clustering
  • Neural Network Library
    • Fully-Connected layer
    • Convolutional Layer
    • Residual Layer
    • ...
• Dataset Importer
• Autolearn