He'll get the lights for you.

A Nest-inspired application for the Phillips Hue.

For more information, see How It Works.

./bin/install

• Docker

The application will download and build all necessary Docker containers. You will be prompted to press the 'Link' button on your Hue Bridge to complete the installation.

To start Hubert, run:

./bin/start

To stop Hubert, run:

./bin/stop

Hubert learns from your light usage patterns. It does this by logging the state of your lights every so often and running the data against a ridge regression model. Hubert will only start changing the state of your lights once it has determined enough data has been collected and the error rate (calculate by the mean squared error) is within a certain threshold.

All information is recorded to a Cassandra database. A python cron then reads the information and makes predictions on based on the current conditions.

The first few weeks of data collection are to train the ridge regression model. Hubert does not start altering the state of your lights until enough data has been collected and the mean squared error is within an acceptable limit.

Hubert's algorithm is simple in that it loops over a range of polynomials (0-9) and a range of alphas (0.0, 1e-8, 1e-5, 1e-1) to determine the fit that best minimizes the test error. There are plans to implement k-fold cross-validation or something similar to best fit smaller amounts of data.

Disclaimer: Python and Machine Learning are both very new to me. I always welcome and appreciate constructive feedback.

Hubert is composed of four containers.

1. hubert - The main processor; Sends updates to Cassandra container; Processes RabbitMQ messages to alter the state of the lights
2. learn - Machine learning algorithm; Learns from usage and sends messages to RabbitMQ for processing
3. rabbitmq - A RabbitMQ instance used for publish-subscribe
4. cassandra - A Cassandra instance used for data collection

Hubert applies linear regression to build opinions on the states of your Hue lights. Python and scikit-learn are used to build intelligence from the data the application collects. Hubert is able to determine when your lights should be on or off and will automatically adjust these settings based on feedback given.

All source code for the machine learning script can be found in script/. The docker container learn is setup to be run via a cron.

You can setup a cron by easily adding the following to your crontab

* * * * * ./bin/learn

We recommend running the machine learning algorithm every minute.

Development for Hubert is relatively straight-forward.

• Node.js
• Docker

Use git to checkout the repository:

git checkout https://github.com/tlackemann/hubert.git hubert

Then install the application by running:

cd hubert && npm install && npm run build

This will install all required modules and compile the source code.

In order to use this application, you'll first need to register it with your Hue Bridge. This can easily be done by following the below steps:

1. Press the 'link' button on your Hue Bridge
2. Run HUE_USER=my-cool-username node dist/install.js
3. Update ./config/default.json with your new username

A new user will automatically be created for you. Update the configuration file with the hash it generates, it will look something like this:

033a6feb77750dc770ec4a4487a9e8db

Don't forget to update the configuration file, otherwise the application will not be able to communicate properly.

In addition to creating a bridge user, you'll also need to seed a Cassandra instance with the hubert schemas.

To seed the Cassandra instance, start the Cassandra container by running:

docker-compose up cassandra

Then run the following to seed the necessary schemas:

docker run -it --link hubert_cassandra_1:cassandra --rm hubert_cassandra sh -c 'exec cqlsh "$CASSANDRA_PORT_9042_TCP_ADDR" -f /docker/hubert.cql'

Finally, build the machine learning container so that we can execute it as a cron.

docker build -t hubert_learn:latest script/

Now add the following to your crontab (make sure to update /path/to/hubert with the correct folder):

* * * * * /path/to/hubert/bin/learn

Start the application by running:

docker-compose up

This will start all containers necessary to run Hubert.

The development Docker container for the Node.js application uses nodemon and Babel to compile and reboot the instance whenever a file is changed. Since it's faster to run ES5 code (currently) with Node, we compile ES6 down to ES5 automatically when a file change is detected. The development version maps the root directory to the application directory in the Docker instance. Any changes made locally will instantly be available on the Docker container.

All configuration can be found in config/default.json. This project utilizes config which allows you to override any configuration setting easily by creating an environment-specific .json file such as config/production.json.

Experimental: Hubert integrates with Twilio to alert you when the state of your light has changed. To enable Twilio, simply update the configuration file with your Twilio credentials.

Copyright 2016 Thomas Lackemann

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

  http://www.apache.org/licenses/LICENSE-2.0

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
See the License for the specific language governing permissions and
limitations under the License.