estpars: estpars
In tsiR: An Implementation of the TSIR Model

This function computes the set up to run the TSIR model, i.e. reconstructs susecptibles and estimates beta and alpha. This can be plugged into simulatetsir.

estpars(
  data,
  xreg = "cumcases",
  IP = 2,
  seasonality = "standard",
  regtype = "gaussian",
  sigmamax = 3,
  family = "gaussian",
  link = "identity",
  userYhat = numeric(),
  alpha = NULL,
  sbar = NULL,
  printon = F
)

`data`	The data frame containing cases and interpolated births and populations.
`xreg`	The x-axis for the regression. Options are 'cumcases' and 'cumbirths'. Defaults to 'cumcases'.
`IP`	The infectious period in weeks. This should be the same as your timestep. Defaults to 2 weeks.
`seasonality`	The type of contact to use. Options are standard for 52/IP point contact or schoolterm for just a two point on off contact or none for a single contact parameter. Defaults to standard.
`regtype`	The type of regression used in susceptible reconstruction. Options are 'gaussian', 'lm' (linear model), 'spline' (smooth.spline with 2.5 degrees freedom), 'lowess' (with f = 2/3, iter = 1), 'loess' (degree 1), and 'user' which is just a user inputed vector. Defaults to 'gaussian' and if that fails then defaults to loess.
`sigmamax`	The inverse kernal width for the gaussian regression. Default is 3. Smaller, stochastic outbreaks tend to need a lower sigma.
`family`	The family in the GLM regression. One can use any of the GLM ones, but the options are essentially 'poisson' (with link='log'), 'gaussian' (with link='log' or 'identity'), or 'quasipoisson' (with link='log'). Default is 'gaussian'.
`link`	The link function used with the glm family. Options are link='log' or 'identity'. Default is 'identity'. to include some bayesian approaches. For 'bayesglm' we use a gaussian prior with mean 1e-4.
`userYhat`	The inputed regression vector if regtype='user'. Defaults to NULL.
`alpha`	The mixing parameter. Defaults to NULL, i.e. the function estimates alpha.
`sbar`	The mean number of susceptibles. Defaults to NULL, i.e. the function estimates sbar.
`printon`	Whether to show diagnostic prints or not, defaults to FALSE.

## Not run: 
require(kernlab)
London <- twentymeas[["London"]]
parms <- estpars(London)
names(parms)
sim <- simulatetsir(London,parms=parms,inits.fit=FALSE)
plotres(sim)

## End(Not run)