tf_linear_regression.R

library(tensorflow)

tf$reset_default_graph()

x_data <- runif(100, min=0, max=1)
y_data <- x_data * 0.1 + 0.3 + rnorm(n, mean=0, sd=0.025)

W <- tf$Variable(tf$random_uniform(shape(1L), -1.0, 1.0))
b <- tf$Variable(tf$zeros(shape(1L)))
y <- W * x_data + b

lmfit <- lm(y_data~x_data) 
plot(x_data, y_data)
abline(a=lmfit$coefficients[1], b=lmfit$coefficients[2],col="green")

# Minimize the mean squared errors.
loss <- tf$reduce_mean((y - y_data) ^ 2)
optimizer <- tf$train$GradientDescentOptimizer(0.5)
train <- optimizer$minimize(loss)

# Launch the graph and initialize the variables.
sess = tf$Session()
sess$run(tf$initialize_all_variables())

# Fit the line (Learns best fit is W: 0.1, b: 0.3)
for (step in 1:201) {
  sess$run(train)
  if (step %% 20 == 0)
    cat(step, "-", sess$run(W), sess$run(b), "\n")
  b_ <- sess$run(b)
  W_ <- sess$run(W)
  plot(x_data,y_data)
  abline(a=lmfit$coefficients[1], b=lmfit$coefficients[2],col="green")
  lines(x_data, b_+W_*x_data, col="red")
  Sys.sleep(0.1)
}