Stochastic_Newton_Algo.py

import numpy as np
import pandas as pd
import sympy as sym


def stochastic_newton_algorithm(dataframe: pd.DataFrame, n=10):
    d = len(dataframe.columns)
    thetas_init = np.ones(d + 1)
    S_inv_init = np.identity(d + 1)
    # thetas_init = [sym.Symbol("theta0")]
    vectors = []

    for k in list(dataframe):
        # thetas/parameters/weights
        # thetas_init.append(sym.Symbol("theta{}".format(k + 1)))
        # self.symbols.append(sym.Symbol("x{}".format(i)))
        # vectors.append(sym.Symbol('x{}'.format(k)))
        # number_column = dataframe.loc[:, k]
        # numbers = number_column.values
        numbers = dataframe[k].to_numpy()
        vectors.append(numbers)
    yVector = vectors[len(vectors)-1]
    xVectors = vectors[:len(vectors)-1]
    xVectors.insert(0, 1)
    # vectors.append(sym.Symbol("y"))
    # number_column = dataframe.loc[:, d]
    # ynumbers = number_column.values
    # vectors.append(ynumbers)
    print(xVectors)
    xVectors_np = xVectors.flatten()
    print(xVectors_np)
    print(yVector)
    #Change xVectors to vectors
    S_inv = 0
    for i in range(1, n):
        # if i > 1:
        #     None
        # else:
        exponent = np.dot(thetas_init, xVectors)
        print(exponent)
        pi = (sym.exp(exponent)) / (1 + sym.exp(exponent))
        a = pi * (1 - pi)
        S_inv = S_inv_init - a * (1 / (1 + a * np.dot(np.dot(xVectors, S_inv_init), xVectors))) * (
            np.dot(np.dot(np.dot(S_inv_init, xVectors), xVectors), S_inv_init))
        thetas = thetas_init + np.dot(S_inv, xVectors)*(vectors[len(vectors) - 1] - pi)
        print(thetas)


if __name__ == '__main__':
    xdata = [1.6907, 1.7242, 1.7552, 1.7842, 1.8113, 1.8369, 1.8610, 1.8839]
    # ydata = [6, 13, 18, 28, 52, 53, 61, 60]
    ydata = [1, 0, 1, 1, 1, 0, 0, 1]
    data = {'xdata': xdata,
            'ydata': ydata}
    df = pd.DataFrame(data)
    print(df)
    # number_column = df.loc[:, 1:]
    # numbers = number_column.values
    # print(numbers)
    # print(df[0].tolist())
    stochastic_newton_algorithm(df, 10)