regressions 1.0

Generate regression models from data

Regressions Library

A collection of algorithms for fitting data to different functional models by using matrices. This library will be made publically available after it is uploaded to Python's database of libraries. It contains all the code for determining regression equations, as well as the code for evaluating said regressions and presenting their results in a raw format.

Contents

1. File Structure
2. Code Examples

File Structure

regressions_library  
|-- matrices  
|   |-- matrix.py  
|   |-- magnitude.py  
|   |-- dot_product.py  
|   |-- column.py  
|   |-- additions.py  
|   |-- scalar.py  
|   |-- multiplication.py  
|   |-- determinant.py  
|   |-- cofactors.py  
|   |-- minors.py  
|   |-- inverse.py  
|   |-- transpose.py  
|-- regressions  
|   |-- run_all.py  
|   |-- best.py  
|   |-- error.py  
|   |-- linear.py  
|   |-- quadratic.py  
|   |-- cubic.py  
|   |-- hyperbolic.py  
|   |-- exponential.py  
|   |-- logarithmic.py  
|   |-- logistic.py  
|   |-- sinusoidal.py  
|-- tests  
|   |-- matrices.py  
|   |-- regressions.py  
|-- READE.md  
|-- .gitignore  

Code Examples

Linear Regression

def linear(data):
    independent_matrix = []
    dependent_matrix = []
    for i in range(len(data)):
        independent_matrix.append([data[i][0], 1])
        dependent_matrix.append([data[i][1]])
    transposition = transpose(independent_matrix)
    product = multiplication(transposition, independent_matrix)
    product_matrix = matrix(product, dtype='float')
    inversion = inv(product_matrix)
    inversion_list = matrix.tolist(inversion)
    second_product = multiplication(inversion_list, transposition)
    solution = multiplication(second_product, dependent_matrix)
    equation = lambda x: solution[0][0]*x + solution[1][0]
    inaccuracy = error(data, equation)
    result = {
        'constants': solution,
        'error': inaccuracy
    }
    return result

Error Calculation

def error(data, equation):
    summation = 0
    for i in range(len(data)):
        summation += (data[i][1] - equation(data[i][0]))**2
    result = summation**(1/4)
    return result