Predict Housing Prices with Tensorflow and AI Platform - Google Cloud Qwiklabs

Objective

Automate deployment, simplify code for provisioning to achieve speed and accuracy.

Guide

How It Works

Original training file "cloud-ml-housing-prices.ipynb" was simplified (removed markdown and extracted bash commands into bash shell script) and only left with Jupyter notebook executable syntax.
A single bash script was created with extracted bash commands from original training file and were coded systematically to be executed as per lab requirement.

Steps to perform this Qwiklabs

1. Visit lab URL (https://google.qwiklabs.com/focuses/3644?parent=catalog) and login to your Qwiklabs account
2. Launch this lab
3. Open Google Console
4. Accept Google Cloud Platform Terms of Service
5. Activate Cloud Shell
6. On Cloud Shell prompt, type following command and hit enter
   source <(curl -s https://raw.githubusercontent.com/donchai/Predict-Housing-Prices-with-Tensorflow-and-AI-Platform/master/predict.sh)
7. Follow on-screen instructions on Cloud Shell to complete this lab

Screenshot

1. Accept Google Cloud Platform Terms of Service
   
2. Stage 1 "Create Storage Bucket"
   
3. Stage 2 "Create the AI Platform notebook instance"
4. Stage 3 "Download lab notebook"
5. Stage 4 Before Complete "Train and deploy the Model for Predictions"
6. Stage 4 After Complete "Train and deploy the Model for Predictions"

Contribution

Contributors are welcome to participate in this project to enhance the deployment scripts in order to achieve full automation e.g.
-- Interact with Google Console via command line (to obtain username and password for automation after launch lab instead of manually copy and paste)
-- Accepting Google Cloud Platform Terms of Service via command line
-- Optimise the scripts to speed up the deployment time
-- Others

Limitation

Unable to accept Google Cloud Platform Terms of Service from command line, still finding/exploring alternative to automate this manual step.

Workaround

Login to Google Console and accept Google Cloud Platform Terms of Service.

Reference

https://google.qwiklabs.com/focuses/3644?parent=catalog

Remark

1. File extension with .orig refer to original file without modification/simplification for reference purpose, not in use during deployment.
2. File extension with .bak refer to development file for future enhancement reference - not in use during deployment.

Credit

Senpai Gil - The Linux Guru for sharing command tip and trick.

Buy me a coffee