ENLITEN - A dataset and its associated analysis code for the paper entitled "Designing sensor sets for capturing energy events in buildings"

Data collection method:

Details of the methodology may be found in the associated manuscript. All data was collected approximately from Aug 2013 to Oct 2013. An instruction for the participants attached here (instruction.pdf) shows the way how the data was collected in the individual homes.

Technical details and requirements:

The data files are in comma-separated (.csv) and tab-separated (.txt) plain text formats. The .py source code files target Python v2.x and use only standard libraries. The R source code files use the following libraries available from CRAN: entropy, ggplot2, plyr, randomForest, reshape2, scales. Some pre-compiled visualisations are included as PDF files.

Additional information:

The zip file contains a set of top level data and source code files, a further files arranged into 5 directories. The 'analysis_datasets' directory contains the main dataset used for the paper. It also includes the machine learning algorithm code (random forest in this case), its associated data and a document showing a selected feature set. The 'gt', 'sd', and 'tf' directories contain raw data collected by one of participants. The 'set_viz' directory contains visualisation code as well as knapsack (cost–benefit) optimisation code for the selection of minimal sensor sets subject to the budget.