Skip to contents

quickfit: Toolbox of model fitting helper functions

quickfit was intended to be an R package to help with simple model fitting tasks in epidemiology and as a central place to store helper functions used in Epiverse-TRACE.

The development of quickfit has been suspended it is no longer considered necessary to have a dedicated package within Epiverse-TRACE to conduct model fitting, and helper functions will remain in the package they were developed in and shared/copied directly across packages rather than requiring taking on a dependency to import them. Development may resume if the need for a utility package becomes apparent.

quickfit was developed at the Centre for the Mathematical Modelling of Infectious Diseases at the London School of Hygiene and Tropical Medicine as part of the Epiverse Initiative.

Installation

You can install the development version of quickfit from GitHub with:

# check whether {pak} is installed
if(!require("pak")) install.packages("pak")
pak::pak("epiverse-trace/quickfit")

Quick start

The examples below show the existing functionality; this is not currently planned to be developed further.

These examples illustrate some of the current functionalities:

Generate some simulated data, define a likelihood, then estimate MLE, or MLE and 95% confidence interval based on profile likelihood:

sim_data <- rnorm(50, 4, 2)

# Define likelihood function
log_l <- function(x,a,b) dnorm(x, a, b, log = TRUE)

# Estimate MLE
estimate_mle(log_l, sim_data, n_param = 2, a_initial = 3, b_initial = 1)
#> $estimate
#>        a        b 
#> 3.896134 1.987079 
#> 
#> $log_likelihood
#> [1] -105.2867

# Estimate 95% CI based on profile likelihood
calculate_profile(
  log_l, 
  data_in = sim_data, 
  n_param = 2, 
  a_initial = 3, 
  b_initial = 1, 
  precision = 0.01
)
#> $estimate
#>        a        b 
#> 3.896134 1.987079 
#> 
#> $profile_out
#>       a1       a2       b1       b2 
#> 3.331714 4.461714 1.651760 2.456527

Additionally, multiple distribution models can be compared (for censored and non-censored data).

multi_fitdist(
  data = rlnorm(n = 100, meanlog = 1, sdlog = 1), 
  models = c("lnorm", "gamma", "weibull"), 
  func = fitdistrplus::fitdist
)
#>    models    loglik      aic      bic
#> 1   lnorm -243.5509 491.1018 496.3121
#> 2 weibull -255.4449 514.8898 520.1001
#> 3   gamma -255.5247 515.0494 520.2597

Help

To report a bug please open an issue; please note that development on quickfit has been suspended, therefore it is not guaranteed that all issues will be responded to.

Contributions

Contributions are welcome via pull requests.

Development on quickfit has been suspended.

However, if you think this package could be developed for a specific use case then contributions are very welcome as issues, or on the main Epiverse Discussion board.

Code of Conduct

Please note that the quickfit project is released with a Contributor Code of Conduct. By contributing to this project, you agree to abide by its terms.