How to prepare dataset for a tf.estimator model?

Question

import numpy as np
import pandas as pd
import tensorflow as tf
from sklearn.model_selection import train_test_split
wine=pd.read_csv(r"C:\\Users\\subha\\Documen