如何在R语言mgcv包gam函数中正确传递自定义weights vector

问题根源 #

mgcv包的gam函数在解析weights参数时，默认会在传入的data数据框或全局环境中查找对应变量，而非直接使用自定义函数的局部参数。当你在check_mod里传入weights_vector时，这个变量是函数的局部对象，gam的求值机制找不到它，因此抛出object 'weights_vector' not found错误。

解决方案 #

下面提供两种无需依赖全局环境变量名的可行方法：

方法1：临时将权重向量加入数据框

把权重向量临时添加到传入的data中，用完后再移除，避免污染原数据：

check_mod <- function(formula, dist, data, weights_vector = NULL, validation_data = NULL) {
  # 保存原数据列名，用于后续清理
  original_cols <- colnames(data)
  temp_weight_col <- ".temp_weights"
  
  if (!is.null(weights_vector)) {
    # 临时添加权重列到数据框
    data[[temp_weight_col]] <- weights_vector
    # gam中引用临时列
    model <- gam(formula, family = dist, data = data, method = "ML", weights = .temp_weights)
    # 移除临时列，恢复原数据结构
    data <- data[, original_cols, drop = FALSE]
  } else {
    model <- gam(formula, family = dist, data = data, method = "ML")
  }
  
  prediction_data <- if (!is.null(validation_data)) validation_data else data
  predicted <- predict(model, newdata = prediction_data)
  
  rmse <- sqrt(mean((prediction_data$Chl - predicted)^2))
  
  model_summary <- summary(model)
  
  deviance_explained <- model_summary$dev.expl
  degrees_of_freedom <- sum(model_summary$edf)
  ML <- model_summary[["sp.criterion"]][["ML"]]
  
  return(list(
    aic = AIC(model), 
    rmse = rmse, 
    dev_expl = deviance_explained, 
    edf = degrees_of_freedom, 
    ML = ML,
    model = model
  ))
}

方法2：使用do.call传递参数

通过do.call构建参数列表并调用gam，绕过gam的表达式求值逻辑，直接传递权重向量：

check_mod <- function(formula, dist, data, weights_vector = NULL, validation_data = NULL) {
  # 构建gam的基础参数列表
  gam_args <- list(
    formula = formula,
    family = dist,
    data = data,
    method = "ML"
  )
  
  # 如果有权重，添加到参数列表
  if (!is.null(weights_vector)) {
    gam_args$weights <- weights_vector
  }
  
  # 用do.call调用gam
  model <- do.call(gam, gam_args)
  
  prediction_data <- if (!is.null(validation_data)) validation_data else data
  predicted <- predict(model, newdata = prediction_data)
  
  rmse <- sqrt(mean((prediction_data$Chl - predicted)^2))
  
  model_summary <- summary(model)
  
  deviance_explained <- model_summary$dev.expl
  degrees_of_freedom <- sum(model_summary$edf)
  ML <- model_summary[["sp.criterion"]][["ML"]]
  
  return(list(
    aic = AIC(model), 
    rmse = rmse, 
    dev_expl = deviance_explained, 
    edf = degrees_of_freedom, 
    ML = ML,
    model = model
  ))
}

验证测试 #

使用你提供的复现代码调用修改后的check_mod，两种方法都能正常运行，不会再出现权重向量找不到的错误：

# 测试方法1或方法2的函数
res1 <- check_mod(formula_example, distribution_example, data_example, weights_vector = weights1)
res2 <- check_mod(formula_example, distribution_example, data_example, weights_vector = data_example$weights_col)

print(res1)
print(res2)

内容的提问来源于stack exchange，提问作者Paco Herrera