I have a DNNLinearCombinedClassifier to predict if an article get sold or not. I need DNN for feature like description and Linear for features like size, category, price, etc. In general it works, but the weight of the price is too low. The price is currently a numeric_column. But then I change the price from 4€, where it got sold to 400€ where it never will be sold the prediction will only change from 0,38 to 0,36.
So my question is: How I can tell Tensorflow to give more weight to the feature column with the price?
Thanks
