IRT models about shinyitemanalysis HOT 5 CLOSED

Přikládám možnou úpravu pomocí mirt balíčku, která celou záložku IRT models výrazně zrychlí. mirt poskytuje podobné obrázky jako ltm, navíc má však dobře naprogramované (rychlé a obecné) funkce k vyhodnocení modelů atd. - viz zdroják

library(difNLR)
library(mirt)
data(GMAT)
data <- GMAT[, 1:20]

### IRT models
###### Rasch model model = 1, itemtype = "Rasch", constrain = NULL
###### 1PL with the fixed discrimination
###### model <- 'F = 1-20
###### CONSTRAIN = (1-20, a1)'
###### nebo
###### model = 1, itemtype = "2PL", constrain = list((1:ncol(data)) + seq(0, (ncol(data) - 1)*3, 3))
###### 2PL model = 1, itemtype = "2PL", constrain = NULL
###### 3PL model = 1, itemtype = "3PL", constrain = NULL
fitR <- mirt(data, model, itemtype = itemtype, constrain = constrain)

# COEF
coef(fitR)

# PLOTS
# expected total score
plot(fitR, type = "score")

# test information function
plot(fitR, type = "info")

# reliability
plot(fitR, type = "rxx")

# test standard errors
plot(fitR, type = "SE")

# item trace lines - characteristic curves multiple figure
plot(fitR, type = "trace")
plot(fitR, type = "trace", facet_items = F)

# item information trace lines - multiple figure
plot(fitR, type = "infotrace")
plot(fitR, type = "infotrace", facet_items = F)

# expected item scoring function - multiple figure
plot(fitR, type = "itemscore")
plot(fitR, type = "itemscore", facet_items = F)

# test information and standard errors - combination
plot(fitR, type = "infoSE")

# factor scores
fs <- fscores(fitR)
sts <- as.vector(scale(apply(data, 1, sum)))

df <- data.frame(fs, sts)

ggplot(df, aes_string("sts", "fs")) +
geom_point(size = 3) +
labs(x = "Standardized total score", y = "Factor score") +
theme_bw() +
theme(text = element_text(size = 14),
plot.title = element_text(face = "bold", vjust = 1.5),
axis.line = element_line(colour = "black"),
panel.grid.major = element_blank(),
panel.grid.minor = element_blank(),
panel.background = element_blank()) +
theme(legend.box.just = "left",
legend.justification = c(1, 0),
legend.position = c(1, 0),
legend.box = "vertical",
legend.key.size = unit(1, "lines"),
legend.text.align = 0,
legend.title.align = 0)

### comparing two models
fit1 <- mirt(data, 1, "Rasch")
fit2 <- mirt(data, 1, "2PL")
anova(fit1, fit2)

### person fit
personfit(fitR)

`

from shinyitemanalysis.

mirt vypada velmi dobre! Bylo by mozne vytvorit jej zatim jako samostatny list?

from shinyitemanalysis.

• mirt přidán jako samostatný list
• models comparison přidáno jako další záložka v IRT models

Co je ještě potřeba?

from shinyitemanalysis.

mirt vypada pekne.

Zatim nam zlobi pridani baliku mirt na Shiny server. Nejake napady? Hlasi to toto:

• installing source package ‘RcppArmadillo’ ...
  ** package ‘RcppArmadillo’ successfully unpacked and MD5 sums checked
  checking whether the C++ compiler works... yes
  checking for C++ compiler default output file name... a.out
  checking for suffix of executables...
  checking whether we are cross compiling... no
  checking for suffix of object files... o
  checking whether we are using the GNU C++ compiler... yes
  checking whether g++ -m64 accepts -g... yes
  checking how to run the C++ preprocessor... g++ -m64 -E
  checking whether we are using the GNU C++ compiler... (cached) yes
  checking whether g++ -m64 accepts -g... (cached) yes
  checking whether g++ version is sufficient... no
  configure: WARNING: Only g++ version 4.6 or greater can be used with RcppArmadillo.
  configure: error: Please use a different compiler.
  ERROR: configuration failed for package ‘RcppArmadillo’
• removing ‘/usr/lib64/R/library/RcppArmadillo’
  ERROR: dependency ‘RcppArmadillo’ is not available for package ‘mirt’

Kompiler prý nemá update.

from shinyitemanalysis.

U mě to funguje bez problémů, zde radí spustit R v režimu R --vanilla a pak až nainstalovat balíčky.

Teď jsem ještě našla debatu vývojářů přímo na GitHubu. Snad to pomůže!

from shinyitemanalysis.