
import tensorflow as tf
import pandas as pd
import numpy as np
from show import show

data = pd.read_csv('test.csv', header=None)
data = np.array(data)
print(data)

X_np = np.array(data[:,[0,1]])
y_np = np.array(data[:,[2]])

[[0.28488 0.52142 1.     ]
 [0.27633 0.21264 1.     ]
 [0.39748 0.31902 1.     ]
 [0.5533  1.      0.     ]
 [0.44274 0.59205 0.     ]
 [0.85176 0.6612  0.     ]
 [0.60436 0.86605 0.     ]]


x = tf.constant( X_np, dtype=tf.float64 )
y = tf.constant( y_np, dtype=tf.float64 )


w_tensor = tf.constant([[-1, 1]], dtype=tf.float64)
b_tensor = tf.constant([[0]], dtype=tf.float64)
w = tf.Variable(w_tensor)
b = tf.Variable(b_tensor)


alpha = 0.1
epochs = 500

wHistory = []
lossHistory = []

def train():

    global w, b
    
    #record forward operations
    with tf.GradientTape() as tape:
        z = x @ tf.transpose(w) + b
        y_hat = tf.math.sigmoid(z)
        loss_points = -y * tf.math.log(y_hat) - (1-y) * tf.math.log(1-y_hat)
        loss = tf.math.reduce_mean(loss_points)

    #backward: compute gradients    
    [dl_dw, dl_db] = tape.gradient(loss, [w, b])    

    #update weights
    w.assign(w - alpha * dl_dw)
    b.assign(b - alpha * dl_db)

    wHistory.append([w.numpy()[0][0],w.numpy()[0][1],b.numpy()[0][0]])     
    lossHistory.append(loss.numpy())    
    
for e in range(epochs):
    train()       
    
show(data, epochs, wHistory, lossHistory,[-1,1])

Logistic regression - tensorflow algorithm¶