The goal of
dsdp is to estimate probability density functions from a data set using a maximum likelihood method. The models of density functions in use are familiar Gaussian or exponential distributions with polynomial correction terms. To find an optimal model, we adopt a grid search for parameters of base functions and degrees of polynomials, together with semidefinite programming for coefficients of polynomials, and then model selection is done by Akaike Information Criterion.
## Install from CRAN install.packages(dsdp)
You can install the development version of
dsdp from this repository:
## Install from github devtools::install_github("tsuchiya-lab/dsdp")
To install from source codes, the user needs an appropriate compiler toolchain, for example, rtools in Windows, to build
dsdp, along with
Please refer to the tutorial and the reference in tsuchiya-lab.github.io/dsdp/.