Energy Efficiency#
Heating load and cooling load of 768 simulated buildings, to be predicted from eight building parameters. The source simulated 12 different building shapes in Ecotect, and the buildings differ in the glazing area, the distribution of the glazing area, the orientation and other parameters. It is a small, clean regression dataset with two targets: all features are numbers and nothing is missing.
Overview#
Loader |
|
Task |
Regression |
Samples |
768 |
Features |
8 |
Targets |
2 continuous targets: |
Missing values |
None |
License |
CC BY 4.0 |
Features#
The features are stored in this order, which is the order X1 to X8 of the source:
relative_compactness, surface_area, wall_area, roof_area,
overall_height, orientation, glazing_area and glazing_area_distribution. All
of them are stored as float64. The names are the descriptions of the variables table of
the source written in lower case with underscores.
orientation and glazing_area_distribution are integers in the source. In the data
file of the source orientation takes the values 2, 3, 4 and 5, and
glazing_area_distribution the values 0 to 5. The source does not say what these numbers
stand for, so they are stored as numbers without category names. The source does not state
the units of the columns, so none are given here.
Targets#
y has two columns, both stored as float64: the heating load (column 0, Y1 of the
source) and the cooling load (column 1, Y2). The names are in Dataset.target_names. The
source says that the aim is to predict each of the two responses from the eight features, and
that the data can also be treated as a multi-class classification problem if the response is
rounded to the nearest integer. Here it is stored as a regression dataset.
Source and citation#
Source: UCI Machine Learning Repository
Obtained from: UCI Machine Learning Repository, dataset 242, downloaded with the ucimlrepo package: all 768 examples with 8 inputs and 2 targets. The ucimlrepo version is recorded in the file.
Creators: Athanasios Tsanas, Angeliki Xifara
Please cite the dataset as:
Tsanas, A. & Xifara, A. (2012). Energy Efficiency [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C51307.
See also:
Tsanas, A., Xifara, A. (2012). Accurate quantitative estimation of energy performance of residential buildings using statistical machine learning tools. Energy and Buildings, vol. 49.
License#
CC BY 4.0, as stated by the source. The license was checked at the source on 2026-10-25. See License and citation.
Changes made#
Converted to the NumPy NPZ format. Features and targets are stored as float64.
The two targets are stored together: y has the shape (768, 2), the first column is the heating load (Y1 of the source) and the second column is the cooling load (Y2).
The columns X1 to X8 are named after their descriptions in the variables table of the source, written in lower case with underscores, for example Glazing Area Distribution (X8) is stored as glazing_area_distribution. The features are stored in the order X1 to X8.
Note
The buildings are 12 different building shapes simulated in Ecotect. They differ in the glazing area, the distribution of the glazing area and the orientation, among other parameters, and 768 building shapes were obtained by simulating various settings of these characteristics.
Note
The source says the aim is to predict each of the two responses from the eight features, and that the data can also be used as a multi-class classification problem if the response is rounded to the nearest integer. This library stores it as a regression dataset.
Note
orientation and glazing_area_distribution are integers in the source. The source does not say what their numbers stand for, so they are stored as its numbers without category names.
Note
The source does not state the units of the columns, so none are given here.
Example#
>>> from pyml_datasets import load_energy_efficiency
>>> data = load_energy_efficiency()
>>> data.X.shape, data.y.shape
((768, 8), (768, 2))
>>> data.target_names
('heating_load', 'cooling_load')
>>> data.has_missing
False
>>> data.y.min(axis=0).round(2).tolist()
[6.01, 10.9]