Implementing logistic regression using TensorFlow

We’ll employ TensorFlow to implement logistic regression, utilizing click prediction as our illustrative example again. We use 90% of the first 100,000 samples for training and the remaining 10% for testing, and assume that X_train_enc, Y_train, X_test_enc, and Y_test contain the correct data:

1. First, we import TensorFlow, transform X_train_enc and X_test_enc into a NumPy array, and cast X_train_enc, Y_train, X_test_enc, and Y_test to float32:

   >>> import tensorflow as tf
   >>> X_train_enc = enc.fit_transform(X_train).toarray().astype('float32')
   >>> X_test_enc = enc.transform(X_test).toarray().astype('float32')
   >>> Y_train = Y_train.astype('float32')
   >>> Y_test = Y_test.astype('float32')
   

In TensorFlow, it’s common to work with data in the form of NumPy arrays. Additionally, TensorFlow operates with float32...