clean_eload: Clean the elaod data

In samirtouzani/GBMbaseline: Gradient boosting machine for building energy modeling

Description

clean_eload This function remove the observations that have negative eload values or values higher or lower than some upper and lower thresholds. The upper threshold is defined as n percent higher than the quantile corresponding to the given upper probability U_ptresh and the lower threshold is defined as n percent lower than the quantile corresponding to the given lower probability L_ptresh divided by n.

Usage

clean_eload(Data, n = 0.2, L_ptresh = 0.005, U_ptresh = 0.995)

Arguments

Data	A dataframe of training or prediction data.
n	An integer that correspond to a multiplicative coefficient that is used to define the thresholding value. The default value is 0.2 which correspond to 20 percent.
L_ptresh	A numeric that correspond to the probability of the lower quantile used for filtering
U_ptresh	A numeric that correspond to the probability of the upper quantile used for filtering

Value

A dataframe that correspond to the cleaned data

samirtouzani/GBMbaseline documentation built on May 29, 2019, 1:02 p.m.