Linear regression

W,b를 조정해서 참에 가까운값으로 예측한다.

 

import tensorflow as tf

import numpy as np

from tensorflow import keras

from tensorflow.keras.optimizers import SGD

# X, Y 데이터

x_train = [1,2,3]

y_train = [1,2,3]

W = tf.Variable(tf.random.normal([1]), name='weight')

b = tf.Variable(tf.random.normal([1]), name='bias')

#cost/loss function

def compute_cost():

    # XW+b

    h = x_train * W + b

    cost=tf.reduce_mean((h-y_train)**2)

    return cost

optimizer=SGD(learning_rate=0.01) #미분 + 알파 기능

for i in range( 2000):

    optimizer.minimize(compute_cost, var_list=[W,b])

print(i, '코스트', compute_cost().numpy)

1999 코스트 <bound method _EagerTensorBase.numpy of <tf.Tensor: shape=(), dtype=float32, numpy=9.3275736e-07>>

W.numpy(),b.numpy()

(array([0.99887836], dtype=float32), array([0.00254972], dtype=float32))