two_tm: Frequentist inference for the two thermal mass model (2TM)

In paulnorthrop/walls: Statistical Inference for the Thermophysical Properties of Walls from In-Situ Surveys

Description

Brief description

Usage

two_tm(data, air = TRUE, init = NULL, tau = 1/12, hetero = TRUE,
  ep = 1e-06, nls_control = NULL, gnls_control = NULL,
  use_gnls = FALSE, ...)

Arguments

data	A dataset in the format of either cwall_east or cwall_north.
air	A logical scalar. Should we use air (TRUE) or surface (FALSE) temperatures?
init	A numeric vector of length 5. Optional initial estimates of the parameters R1, R2, R3, C1 and C2 of the 2TM.
tau	A numeric scalar. Time interval between successive measurements, in hours.
ep	A numeric scalar. Lower bound imposed on the (positive) parameters R1, R2 and C1. Passed to the lower argument of nls.
nls_control	An optional list of control settings to be passed to nls.
gnls_control	An optional list of control settings to be passed to gnls.
use_gnls	A logical scalar. In the hetero = FALSE) case should we use gnls to fit the model (use_gnls == TRUE) or nls (use_gnls == FALSE)? This is provided as a means to check that these two functions give equivalent results.
...	Further arguments to be pased to nls.

Details

Add details

Value

An object of class "nls" returned from nls.

References

Gori, V. (2017) A novel method for the estimation of thermophysical properties of walls from short and seasonally independent in-situ surveys. Doctoral thesis. UCL (University College London). http://discovery.ucl.ac.uk/1568418/

Examples

### cwall_east --------

# Constant error variance
fit1 <- two_tm(data = cwall_east, hetero = FALSE)
fit1
logLik(fit1)
plot(fit1, form = resid(., type = "p") ~ fitted(.) | in_out)

# Different error variances for Qin and Qout
fit2 <- two_tm(data = cwall_east)
fit2
plot(fit2, form = resid(., type = "p") ~ fitted(.) | in_out)
logLik(fit2)

paulnorthrop/walls documentation built on May 15, 2019, 10:02 p.m.