1.DNNClassifier file. I create a DNNClassifier class which is compatible with sklearn's RandomizedSearchCV. In the class, I create a dnn ,and you can use dropout or set batch_norm_momentum. You can change the neurons, the layers, the activation function and so on. In it, I just make the same neurons size in the hidden layers. You can easily change it. 2.DNNRegressor file. just the same as DNNClassifier. Easy to use.
3. In other file that's some messy codes containing some important operations in tensorflow. If you are familier with tensorflow, you may easily know what the codes means, and you can review this knowledge points, and maybe the code just suit little people including me. 4.HaHa!