Ошибка при регистрации только для данных женского пола (попытка согласовать модель Реншоу и Хабермана с помощью StMoMo) - PullRequest
Новая
       57

Ошибка при регистрации только для данных женского пола (попытка согласовать модель Реншоу и Хабермана с помощью StMoMo)

0 голосов
/ 27 ноября 2018

У меня проблемы с подгонкой модели Реншоу и Хабермана к женским канадским данным.Когда я пытаюсь соответствовать модели, эта ошибка появляется. 

Warning in fit.StMoMo(RH, data = CANfemaleinidata, ages.fit = ages.fit, : StMoMo: The model fitting didn't converge.

Полный контекст для этого показан в представ.

Представляет 

    RH <- rh(link = "logit", cohortAgeFun = "1")
    constLC <- function(ax, bx, kt, b0x, gc, wxt, ages){
    c1 <- mean(kt[1, ], na.rm = TRUE)
    c2 <- sum(bx[, 1], na.rm = TRUE)
    list(ax = ax + c1 * bx, bx = bx / c2, kt = c2 * (kt - c1))
    }
    LC <- StMoMo(link = "logit", staticAgeFun = TRUE, periodAgeFun = "NP",
    constFun = constLC)
    canada <- hmd.mx("CAN", "username", "password", "Canada")
    #> Warning in hmd.mx("CAN", "username", "password", "Canada"):
    #> NAs introduced by coercion
    CANfemale <- StMoMoData(canada, series = "female")
    CANfemaleinidata <- central2initial(CANfemale)
    ages.fit <- 55:89
    wxt <- genWeightMat(ages = ages.fit, years = CANfemaleinidata$years,
    clip = 3)
    LCfitfemale <- fit(LC, data = CANfemaleinidata, ages.fit = ages.fit, wxt = wxt)
    #> StMoMo: The following cohorts have been zero weigthed: 1832 1833 1834 1954 1955 1956 
    #> StMoMo: Start fitting with gnm
    #> Initialising
    #> Running start-up iterations..
    #> Running main iterations.....
    #> Done
    #> StMoMo: Finish fitting with gnm
    RHfitfemale <- fit(RH, data = CANfemaleinidata, ages.fit = ages.fit, wxt = wxt,
    start.ax = LCfitfemale$ax, start.bx = LCfitfemale$bx, start.kt = LCfitfemale$kt)
    #> StMoMo: The following cohorts have been zero weigthed: 1832 1833 1834 1954 1955 1956 
    #> StMoMo: Start fitting with gnm
    #> Initialising
    #> Running main iterations....................................................
    #> ...........................................................................
    #> ...........................................................................
    #> ...........................................................................
    #> ...........................................................................
    #> ...........................................................................
    #> .........................................................................
    #> Done
    #> Warning in gnmFit(modelTools = modelTools, y = y, constrain = constrain, : Fitting algorithm has either not converged or converged
    #> to a non-solution of the likelihood equations.
    #> Use exitInfo() for numerical details of last iteration.
    #> StMoMo: Finish fitting with gnm
    #> Warning in fit.StMoMo(RH, data = CANfemaleinidata, ages.fit = ages.fit, : StMoMo: The model fitting didn't converge.

Что особенно странно в этой ошибке, так это то, что она не возникает, когда я использую данные канадских мужчин (см. Рисунок ниже) 

CANmale <- StMoMoData(canada, series = "male")
CANmaleinidata <- central2initial(CANmale)
ages.fit <- 55:89
wxt <- genWeightMat(ages = ages.fit, years = CANmaleinidata$years,
clip = 3)
LCfitmale <- fit(LC, data = CANmaleinidata, ages.fit = ages.fit, wxt = wxt)
#> StMoMo: The following cohorts have been zero weigthed: 1832 1833 1834 1954 1955 1956 
#> StMoMo: Start fitting with gnm
#> Initialising
#> Running start-up iterations..
#> Running main iterations......
#> Done
#> StMoMo: Finish fitting with gnm
RHfitmale <- fit(RH, data = CANmaleinidata, ages.fit = ages.fit, wxt = wxt,
start.ax = LCfitmale$ax, start.bx = LCfitmale$bx, start.kt = LCfitmale$kt)
#> StMoMo: The following cohorts have been zero weigthed: 1832 1833 1834 1954 1955 1956 
#> StMoMo: Start fitting with gnm
#> Initialising
#> Running main iterations.........................
#> Done
#> StMoMo: Finish fitting with gnm

Как мне обойти эту проблему?

...