#Machine Learning for home automation

This repository contains R scripts to create rules for a house.

##Code

All code uses dplyr data frames. The naming convention is to use lowercase names with dots for functions and function arguments. For variables, Snake case is used.

###generator.R This file contains one function (house.generate.rules) that integrates all other files in order to learn rules.

###db.R This file contains functions to connect, save and load data from the database.

###clean/clean_util.R This file contains utility functions to clean the data: smoothing, sampling, clustering, timestamp transformation.

###clean/lightswitch.R Specific cleaning for light problem.

###train.R This file actually does the learning and interpretation of learning algorithm output

###util.R Generic utility functions, like extracting uniqueId from a data.frame row

###run.R Start rServe that can receive calls from the clojure server

##Dependencies In this project we used the following R libraries:

• mongolite
• jsonlite
• dplyr
• rpart
• rpart.utils
• rpart.plot
• lazyeval
• Rserve

##How to run

1. Make sure you have the database with data and nodes (either using the house_preparing project or running a server instance)
2. Start R
3. Make sure you install all the dependencies in your R environment with install.packages
4. Quit R
5. Set your working directory to the root folder from this project with setwd
6. Execute: Rscript run.R to start Rserve