Combined Cycle Power Plant#

Net hourly electrical energy output of a combined cycle power plant, to be predicted from four hourly average ambient variables. The 9568 data points were collected over 6 years (2006-2011), while the plant was set to work with full load. It is a clean regression dataset: all features are numbers, nothing is missing, and it is the largest regression dataset of the package.

Overview#

Loader

load_combined_cycle_power_plant()

Task

Regression

Samples

9568

Features

4

Targets

1 continuous target, net_hourly_electrical_energy_output

Missing values

None

License

CC BY 4.0

Features#

The features are stored in this order, which is the order of the variables table of the source: temperature (AT, in degrees Celsius), exhaust_vacuum (V, in cm Hg), ambient_pressure (AP, in millibar) and relative_humidity (RH, in percent). All of them are stored as float64. They are hourly averages of ambient variables, taken from sensors around the plant that record them every second, and the source gives them without normalization. According to the source the exhaust vacuum is collected from and has an effect on the steam turbine, and the other three variables affect the performance of the gas turbine.

The relative humidity goes above 100 percent in the source, up to 100.16, and is stored as it is.

Target#

y is the net hourly electrical energy output in MW, stored as float64. The source documents the range 420.26 to 495.76.

Order of the rows#

To allow 5x2 fold statistical tests, the source provides the data shuffled five times, one shuffling in each of the five sheets of its file. The rows of this dataset are in the order of the table that UCI serves through ucimlrepo, whose first row is the first row of the first sheet.

Source and citation#

Please cite the dataset as:

Tfekci, P. & Kaya, H. (2014). Combined Cycle Power Plant [Dataset]. UCI Machine Learning Repository. https://doi.org/10.24432/C5002N.

See also:

  • Tüfekci, P. (2014). Prediction of full load electrical power output of a base load operated combined cycle power plant using machine learning methods. International Journal of Electrical Power & Energy Systems, Volume 60, September 2014, Pages 126-140. https://doi.org/10.1016/j.ijepes.2014.02.027

  • Kaya, H., Tüfekci, P., Gürgen, S. F. (2012). Local and Global Learning Methods for Predicting Power of a Combined Gas & Steam Turbine. Proceedings of the International Conference on Emerging Trends in Computer and Electronics Engineering ICETCEE 2012, pp. 13-18, Dubai.

License#

CC BY 4.0, as stated by the source. The license was checked at the source on 2026-10-26. See License and citation.

Changes made#

  • Converted to the NumPy NPZ format. Features and the target are stored as float64.

  • The columns are named after their descriptions in the variables table of the source, written in lower case with underscores: AT is stored as temperature, V as exhaust_vacuum, AP as ambient_pressure, RH as relative_humidity and the target PE as net_hourly_electrical_energy_output. The features are stored in the order AT, V, AP, RH.

Note

The features are hourly averages of ambient variables. The averages are taken from various sensors located around the plant that record the variables every second. The source gives the variables without normalization.

Note

Units: the temperature is in degrees Celsius, the exhaust vacuum in cm Hg, the ambient pressure in millibar, the relative humidity in percent and the net hourly electrical energy output in MW. The source documents the ranges 1.81 to 37.11 for the temperature, 25.36 to 81.56 for the exhaust vacuum, 992.89 to 1033.30 for the ambient pressure, 25.56 to 100.16 for the relative humidity and 420.26 to 495.76 for the output. The relative humidity goes above 100 percent in the source and is stored as it is.

Note

According to the source the exhaust vacuum is collected from and has an effect on the steam turbine, and the other three variables affect the performance of the gas turbine.

Note

To allow 5x2 fold statistical tests, the source provides the data shuffled five times. This dataset holds the rows in the order of the table that UCI serves through ucimlrepo.

Note

UCI prints the citation, and the name of the first creator on its page, without the letters of the name Tüfekci that are not in the English alphabet. The citation is given here as UCI prints it, the name of the creator as it is written in the papers of the source.

Example#

>>> from pyml_datasets import load_combined_cycle_power_plant
>>> data = load_combined_cycle_power_plant()
>>> data.X.shape, data.y.shape
((9568, 4), (9568,))
>>> data.feature_names
('temperature', 'exhaust_vacuum', 'ambient_pressure', 'relative_humidity')
>>> data.has_missing
False
>>> round(float(data.y.min()), 2), round(float(data.y.max()), 2)
(420.26, 495.76)