splitLexisDT: added warning when there are NA values in the splitting time scale; see issue #140

Author: WetRobot

DESCRIPTION

@@ -1,8 +1,8 @@
 Package: popEpi
 Title: Functions for Epidemiological Analysis using Population Data
 Authors@R: c(person("Joonas", "Miettinen", , "[email protected]", c("aut", "cre")), person("Matti", "Rantanen", , "[email protected]", "aut"), person("Karri", "Seppa", , "[email protected]", "ctb") )
-Version: 0.4.3.19
-Date: 2017-11-08
+Version: 0.4.3.20
+Date: 2017-11-13
 Maintainer: Joonas Miettinen <[email protected]>
 Description: Enables computation of epidemiological statistics where e.g. 
     counts or mortality rates of the reference population are used. Currently 

NEWS

@@ -2,6 +2,7 @@ popEpi 0.4.4
 =====
 * splitLexisDT/splitMulti bug fix: splitting along multiple time scales _sometimes_ produced duplicate transitions (e.g. alive -> dead in the last two rows). see https://github.com/WetRobot/popEpi/issues/138 for details.
 * splitLexisDT/splitMulti now retain time.since attribute; this attribute plays a role in cutLexis
+* known issue: splitLexisDT/splitMulti not guaranteed to work identically to splitLexis from Epi when there are NA values in the time scale one is splitting along.
 
 
 popEpi 0.4.3

NEWS.Rmd

@@ -19,6 +19,7 @@ devtools::load_all()
 
 * splitLexisDT/splitMulti bug fix: splitting along multiple time scales _sometimes_ produced duplicate transitions (e.g. alive -> dead in the last two rows). see https://github.com/WetRobot/popEpi/issues/138 for details.
 * splitLexisDT/splitMulti now retain time.since attribute; this attribute plays a role in cutLexis
+* known issue: splitLexisDT/splitMulti not guaranteed to work identically to splitLexis from Epi when there are NA values in the time scale one is splitting along.
 
 # Changes in 0.4.3
 

NEWS.md

@@ -2,6 +2,8 @@ Changes in 0.4.4
 ================
 
 -   splitLexisDT/splitMulti bug fix: splitting along multiple time scales *sometimes* produced duplicate transitions (e.g. alive -&gt; dead in the last two rows). see <https://github.com/WetRobot/popEpi/issues/138> for details.
+-   splitLexisDT/splitMulti now retain time.since attribute; this attribute plays a role in cutLexis
+-   known issue: splitLexisDT/splitMulti not guaranteed to work identically to splitLexis from Epi when there are NA values in the time scale one is splitting along.
 
 Changes in 0.4.3
 ================

R/splitLexisDT.R

@@ -154,6 +154,11 @@ splitLexisDT <- function(lex, breaks, timeScale, merge = TRUE, drop = TRUE) {
     ts_is_na <- is.na(lex[[timeScale]])
     ts_any_na <- any(ts_is_na)
     if (ts_any_na) {
+      warning("NA values in the time scale you are splitting along ('", 
+              timeScale,"'). Results may deviate from that produced by ",
+              "splitLexis from package Epi. For safety you may want to split ",
+              "using only the data with no NA values and combine the the split",
+              " data with the NA-valued data using rbind.")
       lex_na <- lex[ts_is_na, ]
       lex <- lex[!ts_is_na, ]
     }