LoginSign Up
← Back to all sparks
X

xpose4

INFRA · APIS
Velocity0.0

Diagnostics for Nonlinear Mixed-Effect Models

github.com/uupharmacometrics/xpose4 ↗

NONMEM's veteran diagnostics package, now moving at about one fix a year

pharmacometricsnonmemmodel-diagnosticscompatibilitymaintenance
◆Current state
xpose4 generates diagnostic plots for nonlinear mixed-effect models fitted in NONMEM, a role it has held since 2014 when five packages were consolidated into one. The three most recent releases are a documentation cross-reference update, a change of the default table-reading method to the slower and more reliable one, and a set of file-parsing bug fixes. No new plot types appear anywhere in the window.
◆Where it's heading
Development now responds to the R ecosystem rather than to pharmacometrics: deprecated dplyr and tibble calls, readr and gam changes, and R documentation standards account for most recent entries. The last release with visible new capability is 4.6.0 in 2017, which extended xpose.VPC() with outlier identification, median-positioned lines and bin rugs. The feed's older tags are also unreliable — versions 4.5.0 through 4.5.3 all carry the same 2014 timestamp, so their order says nothing about sequence.
◆Prediction
Expect continued compatibility maintenance at roughly one release a year, triggered by changes in R or in the packages it reads tables with; nothing in these entries points to new diagnostic capability being planned.

◆Recent moves

  1. 11mo ago

    Documentation cross-references updated to R standards

    A single-line release bringing cross-references to other packages in line with R documentation standards. Nothing about the diagnostics changes, and it is the only release in over three years.

    View source ↗
  2. 4y ago

    Default table reading switched to the 'slow' method

    Changes the default in xpose.data() to the slower table-reading method, trading import speed for reliability across NONMEM table formats. A small change with real effect, since it alters what happens for users who never set the argument.

    View source ↗
  3. 5y ago

    Fixes for -99 row filtering, catab/cotab covariates and CSV import

    Three parsing fixes, the most consequential being that -99 rows were not filtered on all plotting variables — a defect that would quietly distort plots rather than fail loudly. Also drops deprecated dplyr and tibble calls.

    View source ↗
  4. 6y ago

    Adjustable censored-line length in kaplan.plot()

    Adds control over the relative length of censored lines in kaplan.plot() and handles directory paths without a trailing slash, alongside a fix to the classic menu system. Small user-facing additions of the kind that have since stopped appearing.

    View source ↗
  5. 8y ago

    Compliance updates for readr and gam changes

    A one-line release reacting to changes in two upstream packages. Representative of what most of this changelog has consisted of since 2017.

    View source ↗
  6. 9y ago

    VPC outlier identification, median lines and bin rugs

    The last release with substantive new capability: xpose.VPC() gained outlier identification, lines positioned at the median of observed x-values per bin, and an optional rug showing bin locations. Documentation also moved to roxygen and the namespace was narrowed to loading only lattice.

    View source ↗