import numpy as np

X = np.array(([2, 9], [1, 5], [3, 6]), dtype='float')
Y = np.array(([92], [86], [89]), dtype = 'float')
X = X / np.amax(X , axis=0)
Y = Y / 100

epochs = 1000
learning_rate = 0.6
inputLayers = 2
hiddenLayers = 3 
outputLayers = 1

wh = np.random.uniform(size = (inputLayers, hiddenLayers))
bh = np.random.uniform(size = (1, hiddenLayers))
w0 = np.random.uniform(size = (hiddenLayers, outputLayers))
b0 = np.random.uniform(size = (1, outputLayers))

def sigmoid(z) :
    return 1 / (1 + np.exp(-z))

def derivative(x) :
    return x * (1-x)

for i in range(epochs) :
    # Forward Propagation
    z_h = np.dot(X, wh) + bh
    sigmoid_h = sigmoid(z_h)
    z_0 = np.dot(sigmoid_h, w0) + b0
    output = sigmoid(z_0)
    # Backward Propagation
    deltaK = (Y - output) * derivative(output)
    deltaH = deltaK.dot(w0.T) * derivative(sigmoid_h)
    w0 = w0 + learning_rate * sigmoid_h.T.dot(deltaK)
    wh = wh + learning_rate * X.T.dot(deltaH)
print(f"Input:\n {X}")
print(f"Actual Output:\n {Y} ")
print(f"Predicted Output:\n {output}")


# OUTPUT :-
# Input:
#  [[0.66666667 1.        ]
#  [0.33333333 0.55555556]
#  [1.         0.66666667]]
# Actual Output:
#  [[0.92]
#  [0.86]
#  [0.89]] 
# Predicted Output:
#  [[0.89561426]
#  [0.87785989]
#  [0.89594741]]