thecraigd/Deep_Neural_Net_with_TensorFlow

Machine Learning project utilising TensorFlow to build a deep neural network model.

Neural Networks

Machine Learning training project utilising TensorFlow to build a deep neural network model and make predictions.

Project status - Complete

Introduction

The intention of this project was to build a functioning neural network using TensorFlow, experiment with hidden layers and nodes and see the effect of other hyperparameters in the model.

This was based on examples used in the excellent Machine Learning Crash Course offered by Google.

Methods used

• EDA
• Machine Learning
• Deep Neural Networks
• Loss Function Plotting
• Gradient Descent
• Activation Functions
• Feature Engineering
• Data visualisation

Technologies used

• Jupyter Notebook
• NumPy
• pandas
• TensorFlow
• Keras
• Matplotlib

Data sources

California Housing Dataset based on the 1990 US Census. The data is at the 'block' level, so each record contains data for multiple houses or apartments.

This is a familiar dataset for Machine Learning training, so apologies for the lack of originality!

Sourced from Google.

Getting started

1. Clone this repo (for help see this tutorial).
2. Raw Data is kept in the CSV files in the data folder of this repo.
3. All code is contained within the Jupyter Notebook for this project, stored in the root folder as Neural_Nets.ipynb

Featured Notebooks

• Neural_Nets

• 

Contact

All feedback is warmly received. Craig Dickson can be contacted at craigdoesdata.com.