v2.4

2025-11-16 21:04:27

Author: DSStoffer

NEWS.md

@@ -47,6 +47,8 @@ remotes::install_github("nickpoison/astsa/astsa_build")
 
 > We'll send this version to CRAN when the 2nd edition of [Time Series: A Data Analysis Approach ...](https://www.routledge.com/Time-Series-A-Data-Analysis-Approach-Using-R/Shumway-Stoffer/p/book/9781041031642) is published. The planned publication date is Jan/Feb 2026.
 
+- added `filter` and `lag` from the `stats` package to `astsa` ... this way, if `astsa` is loaded after `dplyr`, these scripts will work as originally intended. It just decreases the chance that `dplyr` is the _pain in the ass_ that it is (also, there is minimal chance `stats` is loaded after `dplyr`)
+
 - added  🆕 `arma.check`  🆕  to check a given model (seasonal ok) for causality, invertibility, and parameter redundancy with option (`redundancy.tol`) to specify how close the inverse roots have to be to report a warning of parameter redundancy or over-parameterization (default is .1).
 
   - in `arma.spec`, using ideas from `arma.check`, added user option `redundancy.tol`.

astsa_2.4.tar.gz

@@ -9,14 +9,15 @@ importFrom("grDevices", "adjustcolor", "col2rgb", "cm.colors", "dev.flush",
               "rgb2hsv", "xy.coords")
 importFrom("stats", "acf", "AIC", "ar", "ar.yw", "arima", "ARMAtoMA", "as.ts", 
               "BIC", "Box.test", "ccf", "coef", "cor", "cov2cor", "df.kernel", 
-              "diffinv", "dnorm", "fft", "filter", "frequency", "is.ts", 
-              "is.tskernel", "kernapply", "kernel", "lag", "lm", "loess", 
+              "diffinv", "dnorm", "fft", "frequency", "is.ts", 
+              "is.tskernel", "kernapply", "kernel", "lm", "loess", 
               "lowess", "mvfft", "na.fail", "na.omit", "na.pass", "naprint", 
               "nextn", "optim", "pacf", "pchisq", "pf", "plot.ts", "ppoints", 
               "predict", "qf", "qnorm", "qqnorm", "qt", "quantile", "resid", 
               "rgamma", "rnorm", "runif", "sd", "spec.ar", "spec.pgram", 
               "spec.taper", "summary.lm", "time", "ts", "ts.intersect", 
-              "ts.plot", "tsp", "var", "vcov")     
+              "ts.plot", "tsp", "var", "vcov")   
+# `filter` and `lag` removed from above because they are now part of `astsa`  
 importFrom("utils", "setTxtProgressBar", "txtProgressBar")
 
 

astsa_build/NAMESPACE

@@ -0,0 +1,66 @@
+#  this is stats::filter 'as.is'
+
+filter <- function(x, filter, method = c("convolution", "recursive"),
+                   sides = 2L, circular = FALSE, init=NULL)
+{
+    method <- match.arg(method)
+    x <- as.ts(x)
+    storage.mode(x) <- "double"
+    xtsp <- tsp(x)
+    n <- as.integer(NROW(x))
+    if (is.na(n)) stop("invalid value of nrow(x)", domain = NA)
+    nser <- NCOL(x)
+    filter <- as.double(filter)
+    nfilt <- as.integer(length(filter))
+    if (is.na(n)) stop("invalid value of length(filter)", domain = NA)
+    if(anyNA(filter)) stop("missing values in 'filter'")
+
+    if(method == "convolution") {
+        if(nfilt > n) stop("'filter' is longer than time series")
+        sides <- as.integer(sides)
+        if(is.na(sides) || (sides != 1L && sides != 2L))
+            stop("argument 'sides' must be 1 or 2")
+        circular <- as.logical(circular)
+        if (is.na(circular)) stop("'circular' must be logical and not NA")
+        if (is.matrix(x)) {
+            y <- matrix(NA, n, nser)
+            for (i in seq_len(nser))
+                y[, i] <- .Call(C_cfilter, x[, i], filter, sides, circular)
+        } else
+            y <- .Call(C_cfilter, x, filter, sides, circular)
+    } else {
+        if(missing(init)) {
+            init <- matrix(0, nfilt, nser)
+        } else {
+            ni <- NROW(init)
+            if(ni != nfilt)
+                stop("length of 'init' must equal length of 'filter'")
+            if(NCOL(init) != 1L && NCOL(init) != nser) {
+                stop(sprintf(ngettext(nser,
+                                      "'init' must have %d column",
+                                      "'init' must have 1 or %d columns",
+                                      domain = "R-stats"),
+                             nser), domain = NA)
+            }
+            if(!is.matrix(init)) dim(init) <- c(nfilt, nser)
+        }
+        ind <- seq_len(nfilt)
+        ## NB: this .Call alters its third argument
+        if (is.matrix(x)) {
+            y <- matrix(NA, n, nser)
+            for (i in seq_len(nser))
+                y[, i] <-
+                    .Call(C_rfilter, x[, i], filter,
+                          c(rev(init[, i]), double(n)))[-ind]
+        } else
+                y <-
+                    .Call(C_rfilter, x, filter,
+                          c(rev(init[, 1L]), double(n)))[-ind]
+
+    }
+#    y <- drop(y)
+    tsp(y) <- xtsp
+    class(y) <- if(nser > 1L) c("mts", "ts") else "ts"
+    y
+}
+

astsa_build/R/filter.R

@@ -0,0 +1,15 @@
+#  this is stats::lag 'as.is'
+
+lag <- function(x, ...) UseMethod("lag")
+
+lag.default <- function(x, k = 1, ...)
+{
+    if(k != round(k)) {
+        k <- round(k)
+        warning("'k' is not an integer")
+    }
+    x <- hasTsp(x)
+    p <- tsp(x)
+    tsp(x) <- p - (k/p[3L]) * c(1, 1, 0)
+    x
+}

astsa_build/R/lag.R

@@ -6,16 +6,17 @@
 		 palette("default")
 		 }
 
+# no need for below because filter and lag are in astsa as of v2.4
 # warn if dplyr is loaded
-.onAttach <- function(libname, pkgname) {
- if ('package:dplyr' %in% search()) {
-warning("\nThe package `dplyr` is loaded and it most likely will corrupt 
-the base scripts `filter` and `lag` that a time series analyst 
-uses frequently. See `?astsa` or `help(astsa)` for remedies. 
-A simple one is to detach it: `detach(package:dplyr)`"
-)
- }
-}
+#.onAttach <- function(libname, pkgname) {
+# if ('package:dplyr' %in% search()) {
+#warning("\nThe package `dplyr` is loaded and it most likely will corrupt 
+#the base scripts `filter` and `lag` that a time series analyst 
+#uses frequently. See `?astsa` or `help(astsa)` for remedies. 
+#A simple one is to detach it: `detach(package:dplyr)`"
+#)
+# }
+#}
 
 
 # below were used for `stats::pairs` in `ar.boot` and `ar.mcmc` and

astsa_build/R/zzz.R

@@ -28,15 +28,15 @@ We do not always push the latest version of the package to CRAN, but [the latest
 
 ---
 
-⛔ __WARNING:__  If loaded, the package `dplyr` may (and probably will) corrupt the base scripts `filter` and `lag` that a time series analyst uses often. An easy fix if you’re analyzing time series (or teaching a class) is to (tell students to) do the following if `dplyr` is going being used:
+⛔ __WARNING:__  If loaded, the package `dplyr` may (and probably will) mask the base scripts `filter` and `lag` that a time series analyst uses often. An easy fix if you’re analyzing time series (or teaching a class) is to (tell students to) do the following if `dplyr` is going being used:
 
 ```r
 # [1] either detach it if it's loaded but no longer needed
 detach(package:dplyr)  
 
 # [2] or fix it yourself when loading dplyr 
 # this is a great idea from https://stackoverflow.com/a/65186251
-library(dplyr, exclude = c("filter", "lag"))   # remove the culprits
+library(dplyr, exclude = c("filter", "lag"))   # load without the culprits
 dlag    = dplyr::lag                           # then correct ... 
 dfilter = dplyr::filter                        # ... the blunders
 #  Now use `dlag` and `dfilter` in dplyr scripts and

astsa_build/README.md

@@ -4,7 +4,7 @@
 Check an ARMA Model for Causality, Invertibility, and Parameter Redundancy
 }
 \description{
-For a give ARMA model (including seasonal models), reports whether the model is causal,
+For a given ARMA model (including seasonal models), reports whether the model is causal,
 invertible, or (approximately) over-parameterized.
 }
 \usage{

astsa_build/man/arma.check.Rd

@@ -21,7 +21,7 @@ LazyData: \tab yes\cr
 \author{
 David Stoffer <stoffer@pitt.edu>
 }
-\section{Warning}{If loaded, the \code{dplyr} package corrupts the base scripts \code{filter} and \code{lag} among other things. In this case, whenever you analyze time series data, we suggest you either:
+\section{Warning}{If loaded after \code{astsa}, the package \code{dplyr} will mask the base scripts \code{filter} and \code{lag} among other things. In this case, whenever you analyze time series data, we suggest you either:
 
 (1) Detach it if it's loaded but not being used: 
 \preformatted{   detach(package:dplyr)

astsa_build/man/astsa-package.Rd

@@ -0,0 +1,69 @@
+% taken from the stats package almost 'as is'
+
+\name{filter}
+\alias{filter}
+\title{Linear Filtering for Time Series}
+\usage{
+filter(x, filter, method = c("convolution", "recursive"),
+       sides = 2, circular = FALSE, init)
+}
+\arguments{
+  \item{x}{a univariate or multivariate time series.}
+
+  \item{filter}{a vector of filter coefficients in reverse time order
+    (as for AR or MA coefficients).}
+
+  \item{method}{Either \code{"convolution"} or \code{"recursive"} (and
+    can be abbreviated). If \code{"convolution"} a moving average is
+    used: if \code{"recursive"} an autoregression is used.}
+
+  \item{sides}{for convolution filters only. If \code{sides = 1} the
+    filter coefficients are for past values only; if \code{sides = 2}
+    they are centered around lag 0.  In this case the length of the
+    filter should be odd, but if it is even, more of the filter
+    is forward in time than backward.}
+
+  \item{circular}{for convolution filters only.  If \code{TRUE}, wrap
+    the filter around the ends of the series, otherwise assume
+    external values are missing (\code{NA}).}
+
+  \item{init}{for recursive filters only. Specifies the initial values
+    of the time series just prior to the start value, in reverse
+    time order. The default is a set of zeros.}
+}
+\description{
+  Applies linear filtering to a univariate time series or to each series
+  separately of a multivariate time series.
+}
+\value{
+  A time series object.
+}
+\details{
+  Missing values are allowed in \code{x} but not in \code{filter}
+  (where they would lead to missing values everywhere in the output).
+
+  Note that the recursive filter gives
+  \deqn{x_t =  \phi_1 x_{t-1} + \cdots + \phi_p x_{t-p} + w_t}{x[t] =
+     phi[1]*x[t-1] + \dots + phi[p]*x[t-p] +  w[t] }
+
+  The convolution filter is
+  \deqn{x_t = \theta_0 w_{t+o} + \cdots + \theta_{q-1} w_{t+o-(q-1)}}{x[t] = theta[0]*w[t+o] + \dots + theta[q]*x[i+o-(q-1)]}
+
+  where \code{o} is the offset: see \code{sides} for how it is determined.
+}
+\note{
+This is \code{stats::filter}. Because of the mess created by \code{dplyr}, now if \code{astsa} is loaded after \code{dplyr}, this will take back the script as originally intended.
+}
+
+\examples{
+# convolution (default)
+wgts = c(.5, rep(1,11), .5)/12
+soif = filter(soi, sides=2, filter=wgts)
+tsplot(cbind(soi, soif), col=2*2:1, lwd=1:2, spaghetti=TRUE)
+
+# recursive
+w = rnorm(300) # 50 extra to avoid startup problems
+x = filter(w, filter=c(1.5, -.75), method="recursive")[-(1:50)]
+tsplot(x, main="autoregression", col=4, gg=TRUE)
+}
+\keyword{ts}