R语言使用nlme包nlsList分组拟合非线性回归报错如何解决

错误原因 #

• 变量名大小写不匹配：数据集内响应变量名为Exp_flux（小写f），nlsList公式中写为Exp_Flux（大写F），R大小写敏感导致变量识别失败，是出现空返回/报错的核心原因。
• 示例数据构造代码存在语法错误：Exp_flux <- 3.5*exp((Ts-10)/10缺少右括号，会导致数据集生成失败。
• 初始值设置适配性不足：当前给定的q=2与模拟数据的真实参数（q≈2.718）存在偏差，若实际数据波动更大，不合适的初始值会导致部分分组拟合不收敛，触发nlsList返回异常。

修正后的nlsList实现代码 #

# 加载依赖包
library(nlme)

# 修正后的数据集构造代码
Date <- as.POSIXct(c("2021-05-25","2021-05-20", "2021-05-21","2021-05-22",
"2021-05-23","2021-05-24" ,"2021-05-25","2021-05-20", "2021-05-21","2021-05-22",
"2021-05-23","2021-05-24" ,"2021-05-25","2021-05-20", "2021-05-21","2021-05-22",
"2021-05-23","2021-05-24" ,"2021-05-25","2021-05-20", "2021-05-21","2021-05-22",
"2021-05-23","2021-05-24" ,"2021-05-25"))
Ts <- rnorm(25, mean=10, sd=0.5)
# 补全右括号
Exp_flux <- 3.5*exp((Ts-10)/10)
Collar <- as.factor(c("t1","t2","t3","t4","t5","t1","t2","t3","t4","t5","t1","t2","t3","t4",
"t5","t1","t2","t3","t4","t5","t1","t2","t3","t4","t5"))
df <- data.frame(Date,Collar,Ts,Exp_flux)

# 修正变量名、调整初始值后的nlsList代码
fit.collars <- nlsList(Exp_flux ~ SRref*q^((Ts-10)/10)| Collar,
                       data = df,  
                       start = list(SRref = 3.5, q = exp(1)), 
                       na.action = na.omit)
# 正常输出所有分组的拟合结果
summary(fit.collars)

更灵活的替代实现方案 #

如果存在部分分组拟合不收敛的情况，可使用嵌套分组+批量拟合的方案，该方案支持自动捕获拟合失败的分组，不会中断整体运行：

library(tidyverse)
library(broom)

df_fit <- df %>%
  # 按分组列嵌套数据
  group_by(Collar) %>%
  nest() %>%
  mutate(
    # 逐组拟合，错误返回NA不中断运行
    fit = map(data, ~tryCatch(
      nls(Exp_flux ~ SRref*q^((Ts-10)/10), data = .x, start = list(SRref=3.5, q=exp(1))),
      error = function(e) NA
    )),
    # 提取拟合参数
    params = map(fit, ~if(all(is.na(.x))) NA else tidy(.x)),
    # 提取拟合统计量
    fit_stats = map(fit, ~if(all(is.na(.x))) NA else glance(.x))
  )

# 导出所有分组的拟合参数
df_fit %>% unnest(params)
# 导出所有分组的拟合统计量
df_fit %>% unnest(fit_stats)

内容的提问来源于stack exchange，提问作者Torben Callesen