同时加载tidyquant与tsDyn包时，tsDyn包NNET函数示例报错求助

Hey there! This is a super common conflict issue when mixing tidyverse-aligned packages like tidyquant with time-series packages that rely on base R's stats functions. Let me break down what's happening and how to fix it:

Why the error occurs #

• The tidyquant package loads functions from the tidyverse (specifically dplyr), which includes its own version of lag().
• This dplyr::lag() function is built to work with vectors, not time-series (ts) objects.
• tsDyn's NNET() function depends on stats::lag() (the base R version) to handle time-series data. When dplyr::lag() takes precedence in your environment, NNET() can't process the ts object correctly—hence the error you're seeing.

Quick Solutions #

1. Explicitly use stats::lag() in your NNET() call #

When running the NNET function, specify that you want to use the base R stats version of lag instead of the dplyr one. Here's how to adjust the example code:

library(tsDyn)
library(tidyquant)

# Load sample data
data(lynx)

# Run NNET with explicit stats::lag
mod <- NNET(lynx, m = 2, lag = stats::lag)

2. Temporarily override lag() for your time-series code #

If you have multiple lines of tsDyn code, use rlang::local_bindings() to temporarily set lag to stats::lag within a code block. This won't affect your use of dplyr::lag outside the block:

library(tsDyn)
library(tidyquant)
library(rlang)

data(lynx)

# Temporarily use stats::lag for this code block
local_bindings(lag = stats::lag) {
  mod <- NNET(lynx, m = 2)
  # Add any other tsDyn code here
}

3. Avoid accidental function overwrites (alternative approach) #

Load tsDyn first, then use tidyquant functions with explicit namespace calls (e.g., tidyquant::tq_get()) instead of relying on loaded functions. This prevents dplyr::lag() from overwriting the base R version entirely, though it's less convenient than the first two methods.

内容的提问来源于stack exchange，提问作者Quinn