Say you follow a tutorial on the tensorflow website for a wide and deep model (https://www.tensorflow.org/tutorials/wide_and_deep)

I create a model based on the US census data to predict whether or not an individual will make more or less than \$50k given a number of features like age, education, profession, etc. I've been able to create the model as well as create a predictor that uses the model just fine. But is there a way to see what features tensorflow is "weighing" more than others? For instance, does it weigh a higher education more than someones age? (i.e. if someone has a PHD and is 26 years old, is the model more likely to say they make more than \$50k vs someone who has an associates degree and is 55 years old?)

I'm using a DNNLinearCombinedClassifier if it matters to this question.

  • $\begingroup$ Tensorflow provides weights of neutral networks. You can take the initial weights and multiply and sum them according to the weights in the following layers $\endgroup$ – keiv.fly Mar 21 '18 at 18:27
  • $\begingroup$ I tried searching for some documentation on what you mean but couldn't find anything. Is there an article that you can provide that explains this? $\endgroup$ – Joshua Terrill Mar 21 '18 at 19:08

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Browse other questions tagged or ask your own question.