fillNASVDSer: SVD to Interpolate the Missing Values in the Time Series Data

In lspatial/sptemUS: Constrained Spatiotemporal Mixed-Effect Models for Exposure Estimation of Air Pollutants

Description

Function to Use SVD to Interpolate the Missing Values in the Time Series Data

Usage

fillNASVDSer(dset, idF, dateF, valF, k)

Arguments

dset	The data frame for time series. Data format: siteid, date, obs dataframe.
idF	The unique location id like siteid.
dateF	The time column name.
valF	The target variable column name.
k	the priciple component, default 1

Details

This function can be used to fill the missing values in time series for many locations.

Value

The data frame similar to the input dset's structure but with filled values.

Examples

#Using the 2014 PM2.5 time series as an example
data("shdSeries2014")
n=nrow(shdSeries2014)
p=0.1 # Set the proportion of missing values
np=as.integer(n*p)
index=sample(n,np)
shdSeries2014missed=shdSeries2014
shdSeries2014missed[index,"obs"]=NA
shdSeries2014filled=fillNASVDSer(shdSeries2014missed,"siteid","date","obs",k=1)

#Exmine the accuracy:
cor(shdSeries2014filled[index,"obs"],shdSeries2014[index,"obs"])
rmse(shdSeries2014filled[index,"obs"],shdSeries2014[index,"obs"])

lspatial/sptemUS documentation built on May 29, 2019, 3:42 a.m.