成功运行r脚本多次后，出现“ lm.fit 0（非NA）情况下的错误”

Question

这是一个难题。 我正在为大型数据集计算线性模型，并使用“ geom_text_repel”将方程式粘贴到图形上。 我成功多次运行了脚本，但是突然开始出现以下错误 ：

lm.fit（x，y，offset = offset，singular.ok = singular.ok，...）中的错误：0（非NA）

这太疯狂了，因为我什么都没做。 我已广泛阅读有关此问题的信息，但尚未找到解决方案。 许多人说这是由于数据集的每一行都有NA，因此缺少协变量（ 从dlply内部调用lm会引发“ 0（非NA）情况”错误[r]和R线性回归问题：lm.fit（x ，y，offset =偏移量，singular.ok = singular.ok，...）和lm.fit中的错误（x，y，offset =偏移量，singular.ok = singular.ok，...）0 non-na情况 ）。 我的数据集很完整：

> apply(ro_aue_SO,1,function(x) sum(is.na(x)))  
40 41 42 43 44 45 46 47 48 49 50 51 52 53 
 0  0  0  0  0  0  0  0  0  0  0  0  0  1

该数据还有其他零个子集，并且仍然存在相同的错误。

我尝试使用na.action = na.omit，对错误使用traceback（）以获得更多见识，我尝试了不同的数据输入方法-无效。 因为错误消息只是在脚本完美运行了几个小时之后出现，所以我想知道这是否是系统问题。 我在OSX 10.12.6上使用RStudio v1.0.153和r 3.4.2。

救命！ 救命！ 救命！ 并预先感谢您。

这是我的（简化）代码：

holes_SO <- read.csv(file = 'data.csv', sep = ",", header = TRUE)
holes_SO$depth <- factor(holes_SO$depth)
ro_aue_SO <- subset(holes_SO, holes_SO$field == "ROA")
ot_slope_SO <- subset(holes_SO, holes_SO$field == "OTS")

#Set-up the empty equation
lm_origin_eqn <- function(m){
  eq <- substitute(italic(y) == b %.% italic(x)*","~~italic(r)^2~"="~r2, 
                   list(b = format(coef(m)[1], digits = 2), 
                        r2 = format(summary(m)$r.squared, digits = 3)))
  as.character(as.expression(eq));                 
}

roa_RTO <- ggplot(data = ro_aue_SO, aes(x = soc_concentration_kg_m3, y = co2_flux_µmol_c_m2_s1, color = depth, shape = depth)) +
  geom_point(size = 3) +
  labs(x = "SOC concentration", y = "CO2 Flux") +
  labs(color="Depth", shape= "Depth") +
  ggtitle(expression('RO Aue, CO'[2]*'')) +
  geom_smooth(aes(color = depth), method=lm, se=FALSE, formula=y~x-1, fullrange = TRUE) +
  xlim(-5,45) +
  theme(plot.title = element_text(size = 16, hjust = 0.5, face = "bold"),
        axis.text = element_text(size = 10),
        axis.title = element_text(size = 12)) +
  scale_color_discrete(drop=FALSE) + 
  scale_shape_discrete(drop=FALSE)

#THIS IS WHERE THE ERROR OCCURS
#fill in the linear equation 
  roa_eqns <- ro_aue_SO %>% split(.$depth) %>%
  map(~ lm(co2_flux_µmol_c_m2_s1 ~ soc_concentration_kg_m3 - 1, data = .)) %>%
  map(lm_origin_eqn) %>% 
  do.call(rbind, .) %>%
  as.data.frame() %>%
  set_names("equation") %>%
  mutate(depth = rownames(.))

#paste equations onto graph
roa_RTO_equations <- roa_RTO + geom_text_repel(data = roa_eqns, aes(x = c(0, 0, 0, 0), y = c(125, 115, 105, 95), label = equation),
                                               parse = TRUE, segment.size = 0, show.legend = FALSE)

以及一小部分数据样本（使用“ dput（holes_SO）”生成）：

structure(list(sample_id = structure(c(1L, 2L, 3L, 4L, 10L, 11L, 
12L, 13L, 14L, 15L, 16L, 17L, 18L, 19L, 20L, 21L, 22L, 23L, 24L, 
25L, 26L, 29L, 30L, 31L, 32L, 27L, 28L, 33L, 36L, 37L, 38L, 39L, 
34L, 35L, 5L, 6L, 7L, 8L, 9L, 40L, 41L, 42L, 43L, 44L, 45L, 46L, 
47L, 48L, 49L, 50L, 51L, 52L, 53L), .Label = c("OTS1-0", "OTS1-30", 
"OTS1-60", "OTS1-90", "OTS10-0", "OTS10-20", "OTS10-30", "OTS10-60", 
"OTS10-90", "OTS2-0", "OTS3-0", "OTS3-30", "OTS3-60", "OTS3-90", 
"OTS4-0", "OTS5-0", "OTS5-30", "OTS5-60", "OTS5-90", "OTS6-0", 
"OTS7-0", "OTS7-20", "OTS7-30", "OTS7-60", "OTS7-90", "OTS8-0", 
"OTS8-120A", "OTS8-120B", "OTS8-20", "OTS8-30", "OTS8-60", "OTS8-90", 
"OTS9-0", "OTS9-120A", "OTS9-120B", "OTS9-20", "OTS9-30", "OTS9-60", 
"OTS9-90", "ROA1-0", "ROA1-30", "ROA1-60", "ROA1-90", "ROA2-0", 
"ROA2-30", "ROA3-0", "ROA3-30", "ROA3-60", "ROA3-90", "ROA4-0", 
"ROA4-30", "ROA4-60", "ROA4-90"), class = "factor"), site = structure(c(1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L), .Label = c("OT", "RO"), class = "factor"), field = structure(c(1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
1L, 1L, 1L, 1L, 1L, 1L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 2L, 
2L, 2L, 2L, 2L), .Label = c("OTS", "ROA"), class = "factor"), 
    hole_number = c(1L, 1L, 1L, 1L, 2L, 3L, 3L, 3L, 3L, 4L, 5L, 
    5L, 5L, 5L, 6L, 7L, 7L, 7L, 7L, 7L, 8L, 8L, 8L, 8L, 8L, 8L, 
    8L, 9L, 9L, 9L, 9L, 9L, 9L, 9L, 10L, 10L, 10L, 10L, 10L, 
    1L, 1L, 1L, 1L, 2L, 2L, 3L, 3L, 3L, 3L, 4L, 4L, 4L, 4L), 
    depth = c(0L, 30L, 60L, 90L, 0L, 0L, 30L, 60L, 90L, 0L, 0L, 
    30L, 60L, 90L, 0L, 0L, 20L, 30L, 60L, 90L, 0L, 20L, 30L, 
    60L, 90L, 120L, 120L, 0L, 20L, 30L, 60L, 90L, 120L, 120L, 
    0L, 20L, 30L, 60L, 90L, 0L, 30L, 60L, 90L, 0L, 30L, 0L, 30L, 
    60L, 90L, 0L, 30L, 60L, 90L), co2_flux_µmol_c_m2_s1 = c(1.710293078, 
    0.30924686, 0.36469938, 0.227477037, 1.254479063, 0.752737414, 
    2.257215969, 11.50282226, 3.566654093, 0.69900321, 1.591361818, 
    13.92149665, 22.73002129, 22.45049, 1.109443533, 7.406644295, 
    7.855618003, 17.78010488, 6.471314337, 5.315970134, 6.347455312, 
    11.54719043, 10.11479135, 11.47752926, 2.805488908, 5.222756475, 
    4.377681384, 7.173613131, 14.51864231, 9.729229653, 4.564367185, 
    10.17710718, 7.70956059, 4.382202183, 3.321182297, 3.858269154, 
    7.542932281, 19.88469738, 10.55216436, 3.572542676, 6.530127468, 
    10.78088543, 12.82422246, 3.093747739, 6.956941294, 3.316715055, 
    8.781949843, 7.684561849, 6.142716566, 2.69743231, 9.67046938, 
    7.018872033, 9.475929618), soc_concentration_kg_m3 = c(16.57, 
    1.28, 1.86, 1.63, 16.88, 16.8, 6.59, 5.7, 1.33, 15, 15.67, 
    3.8, 3.95, 3.95, 17.17, 20.5, 21.1, 4.94, 4.27, 2.43, 14.9, 
    16.52, 4.12, 4.59, 4.59, 4.24, 4.24, 15.36, 15.93, 15.93, 
    7.14, 7.14, 3.87, 3.87, 19.21, 20.24, 6.45, 5, 4.85, 40, 
    7.78, 7.78, 3.6, 41.25, 23, 36.67, 23.04, 12.4, 3.33, 35.71, 
    9.66, 12.31, NA)), .Names = c("sample_id", "site", "field", 
"hole_number", "depth", "co2_flux_µmol_c_m2_s1", "soc_concentration_kg_m3"
), class = "data.frame", row.names = c(NA, -53L))

这是我从运行上面的脚本中得到的，并且应该仍然得到的（颜色/标签略有不同）：

Answer 1

是的，您拥有ROA子集中不会出现的某些深度级别，因此这些数据集根本没有观测值。

> holes_SO$depth <- factor(holes_SO$depth)
> ro_aue_SO <- subset(holes_SO, holes_SO$field == "ROA")
> table(ro_aue_SO$depth)

  0  20  30  60  90 120 
  4   0   4   3   3   0 
> ro_split <- split(ro_aue_SO, ro_aue_SO$depth)
> sapply(ro_split, nrow)
  0  20  30  60  90 120 
  4   0   4   3   3   0 
> ms <- lapply(ro_split, function(x) lm(co2_flux_mol_c_m2_s1 ~ soc_concentration_kg_m3 - 1, data = x))
Error in lm.fit(x, y, offset = offset, singular.ok = singular.ok, ...) (from #1) : 
  0 (non-NA) cases

删除这些不会留下任何错误。

ro_aue_SO <- subset(holes_SO, holes_SO$field == "ROA")
ro_aue_SO <- droplevels(ro_aue_SO)
ro_split <- split(ro_aue_SO, ro_aue_SO$depth)
ms <- lapply(ro_split, function(x) lm(co2_flux_mol_c_m2_s1 ~ soc_concentration_kg_m3 - 1, data = x))