clustering_dlc: Daily load curves clustering

In biggproject/biggr: biggr

Daily load curves clustering

Description

Cluster similar daily load curves based on the load curves itself, calendar variables and outdoor temperature

Usage

clustering_dlc(
  data,
  consumptionFeature,
  outdoorTemperatureFeature,
  localTimeZone,
  kMax,
  kMin,
  inputVars,
  loadCurveTransformation,
  nDayParts,
  balanceOutdoorTemperatures,
  nNeighboursAffinity = 7,
  ignoreDates = c(),
  holidaysDates = c(),
  normalisationMethod = "range01"
)

Arguments

data	<data.frame> containing the time series for total energy consumption of a building, the outdoor temperature, or whatever input is needed for clustering.
consumptionFeature	<string> containing the column name the consumption feature in the data argument.
outdoorTemperatureFeature	<string> containing the column name of the outdoor temperature feature in the data argument.
localTimeZone	<string> specifying the local time zone related to the building in analysis. The format of this time zones are defined by the IANA Time Zone Database (https://www.iana.org/time-zones).
kMax	<integer> defining the maximum number of allowed groups in the clustering proceed.
kMin	<integer> defining the minimum number of allowed groups in the clustering proceed.
inputVars	<list of strings> Select a number of features as an input of the clustering. loadCurve: use the transformed version of the consumption daily load curve based on loadCurveTransformation and nDayParts arguments dailyTemperature: use the average daily outdoor temperature dailyConsumption: use the total daily consumption daysOfTheWeek: use a discrete value to represent the days of the week daysWeekend: use a boolean representing whether is weekend or not dailyHolidays: use a boolean representing whether is holiday or not. daysWeek: use a factorial value for each day of the week. loadCurves: loadCurves_columns, month: use the month of the year, dailyMinConsumption: use the minimum consumption of the day dailyMaxConsumption: use the maximum consumption of the day
loadCurveTransformation	<string> that defines the transformation procedure over the consumption load curves. Possible values are: relative: All daily load curves are relative to their total daily consumption. It is the default mode. absolute: The absolute consumption is used to define each daily load curves.
nDayParts	<integer> defining the parts of day used to. Possible values: 24, 12, 8, 6, 4, 3, 2.
ignoreDates	<list of dates> list of dates to ignore (holidays, weather, ..)

Value

<list> dailyClassification <data.frame> in daily frequency, containing the classification of each daily load curve. absoluteLoadCurvesCentroids <matrix> with row names as the identifier of the cluster, and column names as the day hours. This matrix is filled with the hourly average consumption of each cluster detected. clusteringCentroids <matrix> with row names as the identifier of the cluster, and column names as the input variables used by the clustering algorithm. This matrix is filled with the average values of each input variable and cluster. classificationModel <object> containing a simple classification model to predict a load curve based on calendar features. opts: <dictionary> normSpecs: matrix of aggregations used in the Z-score normalisation of the clustering inputs. The different types of aggregations (mean, median, std, ...) are specified as row names, and the different input features are related with column names. loadCurveTransformation: object inputVars: object nDayParts: object

biggproject/biggr documentation built on Oct. 2, 2024, 11:13 p.m.