Using TensorFlow to do regression on a 13 column data set

Question

I would like to do regression on a 13 column data set. The second column is dependent on the rest of the 12 columns. All column contains real number values.

How can I create a neural network using TensorFlow to do the regression? I have tried going through this tutorial but it is too advanced for me.

Thanks in advance for a MWE.

Answer 1

In that tutorial they are using logistic regression, that is a linear binary classifier. They use the class tf.contrib.learn.LinearClassifier as their model.

If you use class tf.contrib.learn.LinearRegressor then you can do linear regression instead of classification.

In that webpage you have tutorials for other models. If you want to create a neural network you have different tutorial in the left menu, for example:

https://www.tensorflow.org/tutorials/mnist/beginners/

In this repository you have python notebooks with the full code of many different neural networks:

https://github.com/tensorflow/tensorflow/tree/master/tensorflow/examples/udacity