How to create feature_columns for numpy matrix input in Tensorflow?

Question

I'm using tensorflow 1.8.0, python 3.6.5. The data is iris data set. Here is the code:

import pandas as pd
import numpy as np
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
import tensorflow as tf

X = iris['data']
y = iris['target']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

input_train=tf.estimator.inputs.numpy_input_fn(x=X_train,
            y=y_train, num_epochs=100, shuffle=False)
classifier_model = tf.estimator.DNNClassifier(hidden_units=[10, 
                  20, 10], n_classes=3, feature_columns=??)

Here is my problem, how do I setup the feature_columns for a numpy matrix?

If I covert the X and y to pandas.DataFrame, I can use the following code for the feature_columns, and it works in the DNNClassifier model.

features = X.columns
feature_columns = [tf.feature_column.numeric_column(key=key) for key in features]

These links might be of use:tf.feature_column docs and tf.estimator.DNNClassifier docs — Andreas Storvik Strauman
– Andreas Storvik Strauman, Commented Jun 30, 2018 at 13:19

Shayan Salehian · Accepted Answer · 2018-12-23 02:59:42Z

1

You can wrap your numpy ndarray in a dictionary and pass it to numpy_input_fn method as input x and then use the key in that dictionary to define your feature_column. Also note that because each data in your X_train has 4 dimensions, you need to specify the shape parameter when defining tf.feature_column.numeric_column. Here is the completed code:

import pandas as pd
import numpy as np
from sklearn.datasets import load_iris
from sklearn.model_selection import train_test_split
import tensorflow as tf

iris = load_iris()

X = iris['data']
y = iris['target']
X_train, X_test, y_train, y_test = train_test_split(X, y, test_size=0.2)

input_train = tf.estimator.inputs.numpy_input_fn(
                  x = {'x': X_train},
                  y = y_train,
                  num_epochs = 100,
                  shuffle = False)

feature_columns = [tf.feature_column.numeric_column(key='x', shape=(X_train.shape[1],))]

classifier_model = tf.estimator.DNNClassifier(
                       hidden_units=[10, 20, 10],
                       n_classes=3,
                       feature_columns=feature_columns)

answered Dec 23, 2018 at 2:59

Shayan Salehian

1493 silver badges16 bronze badges

Sign up to request clarification or add additional context in comments.

2 Comments

aokelly Over a year ago

This isn't working, here's an error I'm getting: "Exception has occurred: AttributeError module 'tensorflow_estimator.python.estimator.api._v2.estimator' has no attribute 'inputs'"

Shayan Salehian Over a year ago

@aokelly This is because you are using tensorflow 2.x. This question and my answer are based on tensorflow 1.8.0.

Collectives™ on Stack Overflow

How to create feature_columns for numpy matrix input in Tensorflow?

1 Answer 1

2 Comments

Your Answer

Linked

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

2 Comments

Your Answer

Sign up or log in

Post as a guest

Linked

Related