Supervised Learning with Python

What does it do?

Implementation in Python of data analysis and machine learning programs utilizing a supervised learning approach over different datasets. The focus was on Perceptron, Multi-Layer Perceptron, and Naive Bayes algorithms.

How to Use?

Clone the Repository

$ git clone https://github.com/alexandreclem/SupervisedLearning-with-Python.git

Dependencies

* keras==2.10.0
* matplotlib==3.6.2
* numpy==1.23.4
* pandas==1.5.1
* pretty_confusion_matrix==0.1.1
* scikit_learn==1.2.1
* seaborn==0.11.2
* tensorflow==2.10.0

• To install the dependencies, use the requirements.txt file present in the project folder.
  • Within the project folder, Run:

    $ pip install -r requirements.txt

Implementations

• Six projects were made involving the already mentioned algorithms. Each directory represents what follows:

  • src/percepetron_gaussian

    • Implementation of the Perceptron algorithm and classification of Gaussian distributions.

  • src/percepetron_ocr

    • Optical character recognition with the Perceptron algorithm.

  • src/mlp_gaussian

    • Multi-Layer Perceptron algorithm for classification of Gaussian distributions.

  • src/mlp_ocr

    • Optical character recognition with the Multi-Layer Perceptron algorithm.

  • src/mlp_time_series

    • MLP to predict daily values of COVID deaths in Brazil, using previous K values (days).
    • Training Period: January to April 2022
    • Testing Period: May 2022

  • src/indoor

    • Comparison of MLP and Naive Bayes algorithms when classifying the floor in an indoor positional system.