Delete previous files (if existing)



In [ ]:

    
# Uncomment as needed

# BigML
#rm bigml.bashrc

# Google
#rm gpred.bashrc
#rm gpred_oauth.json
#rm gpred_service_account.json

BigML

Your BigML username and API key can be found at https://bigml.com/account/apikey. Fill them in in the first two lines below (between the quotation marks.



In [ ]:

    
export user=""
export key="" 

echo "export BIGML_USERNAME=$user" >> bigml.bashrc
echo "export BIGML_API_KEY=$key" >> bigml.bashrc
echo "export BIGML_AUTH=username="$user"\;api_key="$key"" >> bigml.bashrc

Google

Create a new project in the Google Developers Console or use an existing one. Activate billing (enter your Credit Card information, even though there's free quota — the paid subscription is $10/month FYI).

Fill in your project number and your project ID in the first two lines below.



In [ ]:

    
export google_project_number=""
export google_project_id=""
echo "export GPRED_PROJECT_NUMBER=$google_project_number" >> gpred.bashrc
echo "export GPRED_PROJECT_ID=$google_project_id" >> gpred.bashrc

Use link in output from block below to create a new service account and to get the json credentials file:

Click on "Create"
Rename the downloaded file to gpred_service_account.json and put it in this directory



In [ ]:

    
echo "https://console.developers.google.com/project/$google_project_number/apiui/credential/serviceaccount"



In [ ]:

    
echo "Once you have the json credentials file, move it to `pwd`/gpred_service_account.json"

Google Cloud Sorage

All we need to do is to create a boto config file for later use in gspythonlibrary.



In [ ]:

    
FOO="[Credentials]
gs_service_key_file = `pwd`/gpred_service_account.json
[Boto]
https_validate_certificates = True
[GSUtil]
content_language = en
default_api_version = 2
default_project_id = $google_project_id
[OAuth2]
"
printf '%s\n' "$FOO" > ~/.boto
chmod 0400 ~/.boto

Google Prediction API

Use link in output from running block below to activate Google Prediction API in your project.



In [ ]:

    
echo "https://console.developers.google.com/project/$google_project_number/apiui/apis/library"

The line below creates oauth tokens from the service account credentials, which we we will later use to authenticate and initialize the Google Prediction API wrapper.



In [ ]:

    
python create_gpred_oauth.py

View tokens file:



In [ ]:

    
export GPRED_OAUTH_FILE=`pwd`/gpred_oauth.json
echo "export GPRED_OAUTH_FILE=`pwd`/gpred_oauth.json" >> gpred.bashrc
cat $GPRED_OAUTH_FILE

Storing (optional)

For easier configuration on another (virtual?) machine, you can store these files in a private git directory:

bigml.bashrc
gpred.bashrc
gpred_service_account.json
gpred_oauth.json

Add all to bashrc



In [ ]:

    
cat bigml.bashrc >> ~/.bashrc
cat gpred.bashrc >> ~/.bashrc

Go back to the command line and run

source ~/.bashrc

Restart kernel in other Bash notebooks that are already opened, in order to get the .bashrc changes

Add environment variables to jupyter

Using environment variables from jupyter notebooks isn't straightforward because service strips all environment variables but TERM, PATH and LANG. I haven't found a way to automatically change this yet, but the workaround is to edit the script in /etc/init/jupyter.conf to add the contents of bigml.bashrc and gpred.bashrc just before these two lines:

echo 621 > /var/run/ipython_start.pid
exec jupyter-notebook --config='/root/.ipython/profile_nbserver/ipython_config.py'

You should then close all notebooks and restart jupyter with the following command executed as root:

service jupyter restart