GitHub - jcval94/FitUltD: Fit univariate mixed and usual distributions (original) (raw)
FitUltD
The goal of FitUltD is to fit data that can't be fitted with ordinary density functions
Installation
You can install the released version of FitUltD from CRAN with:
install.packages("FitUltD")
Example
This is a basic example which shows you how to fit a multimodal random variable choosing your own distributions:
library(FitUltD) #> Loading required package: mclust #> Package 'mclust' version 5.4.5 #> Type 'citation("mclust")' for citing this R package in publications. #Random Variable set.seed(3110)
RV <- c(rnorm(73,189,12), rweibull(82,401,87), rgamma(90,40,19))
Nombres <- c("norm","weibull","gamma","exp","cauchy")
FIT1 <- FDistUlt(RV, plot = TRUE, subplot = TRUE)
One of the available options is to show the distribution functions that passed the Anderson Darling and Kolmogorov Smirnov tests, as well as their p-value and the proportion of the total distribution.
FIT1[[3]] #> Distribution Dist_Prop Dist AD_p.v KS_p.v #> AD7 gamma(252.339, 293.811)*222.168+0 0.2979592 gamma 0.8859093 0.9397635 #> AD2 norm(86.894, 0.27) 0.3346939 norm 0.5466113 0.7882263 #> AD71 gamma(51.093, 73.537)*2.999+0 0.3673469 gamma 0.4460519 0.6231112 #> estimate1 estimate2 estimateLL1 estimateLL2 method PV_S Obs #> AD7 252.33913 293.8110186 0 222.16825 mge 1.825673 73 #> AD2 86.89350 0.2698051 0 1.00000 mge 1.334838 82 #> AD71 51.09258 73.5366847 0 2.99879 mge 1.069163 90 #> Lim_inf Lim_sup #> AD7 162.249575 222.16825 #> AD2 85.842947 87.33807 #> AD71 1.483556 2.99879
By setting plot
and subplot
arguments as TRUE
, is possible to visualizate each distribution which forms the most accurrate model.
Real data distribution versus fitted model.
Distributions that forms the fitted model.