Skip to content

rishanki/Binary_classification-Keras-ANN-

Repository files navigation

Binary_classification-Keras(ANN)

There are three notebooks in this module and they need to be opened step-wise as you go forward with your study on building your own Binary classification Model using ANN(Artificial Neural Network).

First Notebook: Keras+binary+classifcation+demo.ipynb

Here 2 approaches have been used. One where we have used (Wrappers for the Scikit-Learn API-Keras). Check out the link here https://keras.io/scikit-learn-api/ It is basically creating your own function for the keras model and then calling it using the wrappers. For validation the kfold data split technique is used. The other aproach is our standard sequential initializing module method. Both give an accuracy value which is approximately the same.

Second Notebook: Keras_Demo.ipynb

After you get a hang of the first notebook now you can open the second one. Here we have gone more deep into what all features and factors are available for us to play with in Keras. Numerious test have been done using different parameters to test the performance in all ways. This will give you detailed information.

Third Notebook: Keras_Practice_Session.ipynb

This notebook acts like a guide for your keras journey. Here you have examples, code snippits, theory and hands on practice all put together at one place. Hopefully this will give you comprehensive details on DL using Keras and the help tips in between will atleast initiate a start.

Happy Keras!!

DataSet- Our basic aim is to predict customer churn for a certain bank i.e. which customer is going to leave this bank service. Dataset is small(for learning purpose) and contains 10000 rows with 14 columns. You can download data from https://drive.google.com/file/d/0By9Y49AzZGaUemtpNWtQMWdqRDA/view.