ML, Project

Perceptron Based On Pocket Algorithm

This Python program implements the Perceptron algorithm using the Pocket Algorithm. It is used for binary classification and finding a linear separator that separates two classes of data points. The Pocket Algorithm is employed to find the best linear separator that minimizes the number of misclassifications.

Dependencies

This program uses the following Python libraries:

  • pandas
  • numpy
  • matplotlib

Functions

perceptron_pocket(x_data, y_data, number_of_iteration=300, alpha=0.005)

This function implements the Perceptron algorithm with the Pocket Algorithm to find the best linear separator.

  • x_data: Feature data as a NumPy array or DataFrame.
  • y_data: Target labels as a NumPy array or DataFrame.
  • number_of_iteration: The maximum number of iterations (default: 300).
  • alpha: Learning rate (default: 0.005).

The function returns the best weights and bias that minimize misclassifications.

miss_calc(x_test, y_test, weights, bias)

This function calculates the count of misclassifications on test data using the provided weights and bias.

  • x_test: Feature data for testing as a NumPy array or DataFrame.
  • y_test: Target labels for testing as a NumPy array or DataFrame.
  • weights: Weights obtained from training.
  • bias: Bias obtained from training.

🔗 Link to code