pclm.control: Auxiliary for controlling PCLM fitting
In MaciejDanko/pclm: Penalized Composite Linear Models (PCLMs)

Description Usage Arguments Value Author(s) References See Also

Auxiliary function for controlling PCLM fitting. Use this function to set control parameters of the pclm.default and other related functions.

pclm.control(x.div = 1L, x.auto.trans = TRUE, x.max.ext = 25L,
  zero.class.add = TRUE, zero.class.end = NULL, zero.class.frac = 0.2,
  bs.use = "auto", bs.method = c("MortalitySmooth", "bs"),
  bs.df = c("maxprec", "thumb"), bs.df.max = 250L, bs.deg = 3L,
  opt.method = c("BIC", "AIC"), opt.tol = .Machine$double.eps^0.5,
  pclm.deg = 2L, pclm.max.iter = 100L,
  pclm.lsfit.tol = .Machine$double.eps^0.5,
  pclm.tol = .Machine$double.eps^0.5)

`x.div`	Number of sub-classes within PCLM tim/age class (default is 1). Low value of the parameter makes the PCLM computation faster. It is however recommended to set it to higher value (e.g. 10) for better `nax` estimates.
`x.auto.trans`	Logical indicating if automatically multiple age intervals to remove fractions. `TRUE` is the recommended value. See also examples in `pclm.default`.
`x.max.ext`	Integer defining maximal multiple of an age interval. See also `pclm.interval.multiple`.
`zero.class.add`	Logical indicating if additional zero count class (open interval) should be added after last age class. `TRUE` is the recommended value. See `pclm.nclasses` and `pclm.compmat`.
`zero.class.end`	Positive indicating the end of zero count class = anticipated end of last (open) interval. If set to `NULL` and `zero.class.add == TRUE` then it is calculated automatically based on `zero.class.frac`. See `pclm.nclasses` and `pclm.compmat`.
`zero.class.frac`	Fraction of total range of `x` (age/time vector) added as the last zero-count interval when `zero.class.end == NULL`. Used in `pclm.compmat`. Increase this value if the right tail of the PCLM fit is badly fitted (use `plot.pclm` to diagnose).
`bs.use`	Logical indicating if use B- or P-spline basis to speed-up computations. Possible values: `"auto"`, `TRUE`, and `FALSE`. Used by `pclm.compmat` function.
`bs.method`	Basis for B- or P-spline used by `pclm.compmat` function. Possible values: `"MortalitySmooth"` - gives "P-splines" basis based on `MortSmooth_bbase` of `MortalitySmooth` package (recommended) `"bs"` - gives basic B-splines basis based on `bs` `splines`.
`bs.df`	B- or P- spline degree of freedom (df, number of inner knots) or a way to its calculation used in `pclm.compmat` function. The value is automatically limited by the `bs.df.max`. It can take corresponding values: `"maxprec"` - df equal to the number of ungrouped (raw) age classes (recommended option). `"thumb"` - 'rule of thumb': one knot for the B-spline basis each 4-5 observations. `integer` - df given explicitly.
`bs.df.max`	Maximal number of knots (df) for B- or P-spline basis. Defaut value is 250, but can be decreased if computations are slow. Used in `pclm.compmat`.
`bs.deg`	Degree of the piecewise polynomial for B- or P-spline basis. Default and recommended value is 3. Used in `pclm.compmat`.
`opt.method`	Selection criterion for `lambda` (smooth parameter) in `pclm.opt` function. Possible values are `"AIC"` and `"BIC"` (recommended).
`opt.tol`	Tolerance for `pclm.opt` function that estimates smooth parameter `lambda`.
`pclm.deg`	Order of differences of the components of `b` (PCLM coeficients, beta in the reference [1]). Default value is 2, some other values may cause algorithm to not work properly. Used by `pclm.core` function.
`pclm.max.iter`	Maximal number of iterations in `pclm.core` function. Default is 100, but increase it when got a warning.
`pclm.lsfit.tol`	Tolerance for `lsfit` function used in `pclm.core` function.
`pclm.tol`	Tolerance for `pclm.core` function.