seq.int(0, to0 - from, by) 中的错误：'to' 必须是 R 中的有限数时间序列 [关闭]

Question

我执行时间序列分析。 例如数据

df2=structure(list(supplier = c("TKP", "S7", "Travelfusion", "MyAgent", 
"S7", "TKP", "Travelfusion", "MyAgent", "TKP", "S7", "MyAgent", 
"Travelfusion", "S7", "TKP", "Travelfusion", "MyAgent", "TKP", 
"S7", "Travelfusion", "MyAgent"), date = c("2021-01-06", "2021-01-06", 
"2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", 
"2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", 
"2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", "2021-01-06", 
"2021-01-06", "2021-01-06", "2021-01-06"), hour = c(18L, 18L, 
18L, 18L, 19L, 19L, 19L, 19L, 20L, 20L, 20L, 20L, 21L, 21L, 21L, 
21L, 22L, 22L, 22L, 22L), base_price = c(4770, 49881, 244.45, 
0, 39253, 13168, 101.1, 0, 4156, 12946, 0, 0, 51737, 54711, 0, 
0, 23875, 41853, 52.61, 0)), row.names = c(NA, 20L), class = "data.frame")

我的代码

u_supplier<-unique(df2$supplier)
u_supplier
for(i in 1:length(u_supplier)) {

  s_df<-df[df$supplier==u_supplier[i],]

  date_time <- apply(X = s_df[,c('date','hour')],MARGIN = 1, FUN = function(x) {
    x = paste(as.vector(x),collapse = "")
    return(x)
  })
  dt_index <- seq(from = as.POSIXct(date_time[1],format = "%m-%d-%Y %H"),
                  to  = as.POSIXct(date_time[length(date_time)],format = "%m-%d-%Y %H"),
                  by = "hour")

  ts_data <- xts(x = s_df$base_price, order.by = dt_index)
  names(ts_data)<-c('base_price')

  m<-holt(y = ts_data,h = 24*3)

  ts_data$hlt = m$fitted

  pr_dt_index <- seq(from = as.POSIXct(date_time[length(date_time)],format = "%m-%d-%Y %H")+hours(1),
                     to  = as.POSIXct(date_time[length(date_time)],format = "%m-%d-%Y %H")+hours(24*3),
                     by = "hour")
  pr_s_dt<-cbind(supplier = u_supplier[i],
                 date = gsub(" ","",format(pr_dt_index, "%e/%e/%Y")),
                 hour = hour(pr_dt_index),
                 weekday = as.POSIXlt(pr_dt_index)$wday,
                 base_price = round(m$mean,2))

  write.csv(x = data.frame(pr_s_dt),file = paste0(u_supplier[i],"_results.csv"),row.names = F,quote = F)

  p<-plot(ts_data, main = paste(u_supplier[i],'\nMAE=',round(MAE(y_pred = ts_data$hlt,y_true = ts_data$base_price),2)))
  print(p)
  
}

我不明白为什么会出现这个错误

Error in seq.int(0, to0 - from, by) : 'to' must be a finite number

我在这里看到了类似的主题和解决方案，但这对我没有帮助。 如何修复我的数据中的这个错误？

Answer 1

当您从脚本中的特定行收到错误时，查看该行代码中使用的数据。当您发现您认为的与 R 所说的完全不同时，大多数此类问题都会自行解决（快得多！）。

首先，您的代码中有两个错误：

1. 将 paste(..., collapse="") 替换为 paste(..., collapse=" ")，尤其是因为您特别依赖该空间。
2. 将as.POSIXct 格式修复为"%Y-%m-%d %H"。

---

演练，将df 替换为df2。

在for 循环 (i < 1) 的第一遍，我看到了：

  s_df<-df2[df2$supplier==u_supplier[i],]
  date_time <- apply(X = s_df[,c('date','hour')],MARGIN = 1, FUN = function(x) {
    x = paste(as.vector(x),collapse = "")
    return(x)
  })
  dt_index <- seq(from = as.POSIXct(date_time[1],format = "%m-%d-%Y %H"),
                  to  = as.POSIXct(date_time[length(date_time)],format = "%m-%d-%Y %H"),
                  by = "hour")
# Error in seq.int(0, to0 - from, by) : 'to' must be a finite number

深入了解价值观：

as.POSIXct(date_time[1],format = "%m-%d-%Y %H")
#  1 
# NA 

date_time[1]
#              1 
# "2021-01-0618" 

1. 2021-01-06 不是 "%m-%d-%Y"。
2. 小时前没有空格

试试吧：

date_time <- apply(X = s_df[,c('date','hour')],MARGIN = 1, FUN = function(x) {
  x = paste(as.vector(x), collapse = " ")
  ##                                  ^ add this space
  return(x)
})
as.POSIXct(date_time, format = "%Y-%m-%d %H")
#                         1                         6                         9                        14 
# "2021-01-06 18:00:00 EST" "2021-01-06 19:00:00 EST" "2021-01-06 20:00:00 EST" "2021-01-06 21:00:00 EST" 
#                        17 
# "2021-01-06 22:00:00 EST" 

从那里开始

dt_index <- seq(from = as.POSIXct(date_time[1],format = "%Y-%m-%d %H"),
                to  = as.POSIXct(date_time[length(date_time)],format = "%Y-%m-%d %H"),
                by = "hour")  ##   update two formats   ^^^            ^^^
dt_index
# [1] "2021-01-06 18:00:00 EST" "2021-01-06 19:00:00 EST" "2021-01-06 20:00:00 EST" "2021-01-06 21:00:00 EST"
# [5] "2021-01-06 22:00:00 EST"