在行求值中使用for循环-R版本4.0.0

Question

我读过多个线程解释说，如果有更好的方法我想学习的话，应该不鼓励使用for循环。我会说我已经尝试过将summarize()与group_by()结合使用。

我正在努力实现的是，我想开发一个气候数据库。我已经成功地将R编程为直接从源下载数据，并将列表转换为data.frame。现在我想按月份和年份对多个列进行求和和和/或求平均值。因此，我为什么尝试使用summarize和group_by。我的问题是数据带有代码"M"或"T"，我想保留这些代码，所以我任意地给它们取M=9999和T=9998的整数。我想，当操作代码时，我可以使用for循环逐行求值，将这两个占位符转换为"0"，并返回该子集中的"M"one_answers"T"的数量。

以下是数据到达的方式：

$data
# A tibble: 935 x 8
date             datatype station        value fl_m  fl_q  fl_so fl_t 
<chr>            <chr>    <chr>          <int> <chr> <chr> <chr> <chr>
1 2020-01-01T00:0~ PRCP     GHCND:USW0002~    76 ""    ""    W     "240~
2 2020-01-01T00:0~ SNOW     GHCND:USW0002~     0 "T"   ""    W     ""   
3 2020-01-01T00:0~ SNWD     GHCND:USW0002~     0 "T"   ""    W     ""   
4 2020-01-01T00:0~ TMAX     GHCND:USW0002~    39 ""    ""    W     "240~
5 2020-01-01T00:0~ TMIN     GHCND:USW0002~    -5 ""    ""    W     "240~
6 2020-01-02T00:0~ PRCP     GHCND:USW0002~     3 ""    ""    W     "240~
7 2020-01-02T00:0~ SNOW     GHCND:USW0002~     5 ""    ""    W     ""   
8 2020-01-02T00:0~ SNWD     GHCND:USW0002~     0 ""    ""    W     ""   
9 2020-01-02T00:0~ TMAX     GHCND:USW0002~    11 ""    ""    W     "240~
10 2020-01-02T00:0~ TMIN     GHCND:USW0002~   -10 ""    ""    W     "240~
# ... with 925 more rows

这是我用来把它从列表变成数据的代码。帧：

## Convert a list from NCDC into a data frame
## mso_data is a placeholder file for the downloaded data from NCDC
## mso_light2 is a placeholder for the destination data frame
## NCDC downloads in a list, the data is stored in the $data portion
library(tidyverse)

## first convert from list to data.frame and remove 'station ID' column
mso_light2 <- mso_data$data[, -3]
## remove time from date group
mso_date <- mso_light2[1]
mso_date <- sub("T.*", "", mso_date$date)
mso_light2$date <- mso_date 
## remove flags for fl_so? and fl_t (time)
mso_light2 <- mso_light2[1:5]
## Change 'T' = 9998 & 'M' = 9999
mso_light2$value[mso_light2$fl_m == "T"] <- 9998
mso_light2$value[mso_light2$fl_q == "M"] <- 9999
## pivot data frame
## eventually use to change column names
## v_names <- c('PRCP', 'SNOW', 'SNWD', 'TMAX', 'TMIN')
mso_light2 <- mso_light2[1:3]
mso_light2 <- pivot_wider(mso_light2,
names_from = datatype,  
values_from = value)

这就是数据帧在转换后的样子，我添加了月份和年份以及平均日温度"TAVG"的列：

# A tibble: 187 x 9
# Rowwise: 
date        PRCP  SNOW  SNWD  TMAX  TMIN  TAVG month  year
<date>     <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl> <dbl>
1 2020-01-01    76  9998  9998    39    -5  17       1  2020
2 2020-01-02     3     5     0    11   -10   0.5     1  2020
3 2020-01-03     5     8  9998    61   -38  11.5     1  2020
4 2020-01-04     8  9998     0    33   -66 -16.5     1  2020
5 2020-01-05     5    10     0    33   -21   6       1  2020
6 2020-01-06  9998  9998  9998    33   -38  -2.5     1  2020
7 2020-01-07  9998     0     0    78   -10  34       1  2020
8 2020-01-08     5  9998  9998    44   -27   8.5     1  2020
9 2020-01-09  9998  9998     0     0   -55 -27.5     1  2020
10 2020-01-10     8    10     0   -10   -99 -54.5     1  2020
# ... with 177 more rows

这是我尝试使用summary和group_by:的原始代码

## first format mso_light2$date from <chr> to an actual 'date'
install.packages("chron")
install.packages("openair")
install.packages("lubridate")
library("openair")
library("chron")
library('lubridate')
options(stringAsFactors = FALSE)
mso_light2$date <- as.Date(mso_light2$date, "%Y-%m-%d")
## Turning all daily temperatures into an average
mso_light2 <- mso_light2 %>% rowwise() %>% 
mutate(TAVG = mean(c(TMAX, TMIN), na.rm = T))
## Composing daily data into monthly packages
mso_light2 <- mso_light2 %>%
mutate(month = month(date)) %>%
mutate(year = year(date))
##  mso_PRCP <- mso_light2 %>%
##    group_by(month, year) %>%
##    summarise(PRCP = sum(PRCP)) 
##  mso_SNOW <- mso_light2 %>%
##    group_by(month, year) %>%
##    summarise(SNOW = sum(SNOW)) 
##  mso_TAVG <- mso_light2 %>%
##    group_by(month, year) %>%
##    summarise(TAVG = mean(TAVG)) 
##  summarise(SNOW = sum(SNOW)) %>%
##  summarise(TAVG = mean(TAVG))

问题是我不知道如何删除我的占位符"9999"&amp9998'，并使它们成为'0。所以我一直在尝试开发一个for循环，这就是我所拥有的：

for(i in 1:length(mso_light2$year[[1]])){
startDate <- as.character(mso_light2$date[1])
startDate <- str_split(startDate, "-")
start_year <- startDate[[1]][1]
start_month <- startDate[[1]][2]
start_day <- startDate[[1]][3]

for(j in 1:length(mso_light2$month)){
mso_monthly <- sapply(mso_light2, 
function(x) sum(x[["PRCP"]]), 
use.names = 
paste(start_year, '-', 
start_month, sep = ""))
}

}

请忽略sapply()。我已经尝试了该系列的所有可能功能，但它们都返回错误消息。

这是我不断得到的错误：

FUN(X[[i]]，…(中的错误：未使用的参数(use.names="2020-01"(

sapply只是我在寻求帮助之前尝试的最后一个函数，谢谢。

Answer 1

我知道您正试图从GHCN下载2020年USW00024153站的数据。

library(tidyverse)
dt_path <- "ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily/by_year/2020.csv.gz"
download.file(dt_path, "2020.csv.gz", mode="wb")
#> ID = 11 character station identification code
#> YEAR/MONTH/DAY = 8 character date in YYYYMMDD format (e.g. 19860529 = May 29, 1986)
#> ELEMENT = 4 character indicator of element type 
#> DATA VALUE = 5 character data value for ELEMENT 
#> M-FLAG = 1 character Measurement Flag 
#> Q-FLAG = 1 character Quality Flag 
#> S-FLAG = 1 character Source Flag 
#> OBS-TIME = 4-character time of observation in hour-minute format (i.e. 0700 =7:00 am)
#>  this list ftp://ftp.ncdc.noaa.gov/pub/data/ghcn/daily/by_year/
#>  data dictionary https://www1.ncdc.noaa.gov/pub/data/ghcn/daily/readme.txt

来自这个FTP服务器的数据稍微干净一点，至少日期不包含时间戳。我重用您的列名，因为数据没有标题。还要注意，readr::read_csv()(和data.table::fread()(可以很好地处理压缩文件，因此不需要解压缩。

dt_colnms <- c("station", "date", "datatype", "value", "fl_m", "fl_q", "fl_so", "fl_t")
dt <- readr::read_csv("2020.csv.gz", col_names = dt_colnms, col_types = 'cccdcccc')

数据处理步骤包括：

1. 过滤您需要的站点，忽略数据集中也存在的风柱
2. 透视多个值列(感兴趣的值和标志(
3. 平均温度。既然你只有两列，我看不出有什么理由去rowwise()
4. 从字符日期中提取月和年，并转换日期

dt %>% 
filter(station=="USW00024153", !str_detect(datatype, "^W")) %>% 
pivot_wider(id_cols = "date",
names_from = "datatype",
values_from = c("fl_m", "fl_q","value")) %>% 
mutate(value_TAVG=(value_TAVG+value_TAVG)/2,
month=parse_number(substr(date, 5,6)),
year=parse_number(substr(date, 1,4)),
date=as.Date(date, "%Y%m%d"))

现在，您的最后一步是检查fl_m=="0"的行的替换值；T"；或者其中fl_q＝＝"0"；M〃；。

你本可以在旋转之前完成。然后，转向和总结都将变得更容易：

dt %>% 
filter(station=="USW00024153", !str_detect(datatype, "^W")) %>% 
mutate(value=ifelse(fl_m=="T"&!is.na(fl_m), 0, value),
value=ifelse(fl_q=="M"&!is.na(fl_q), 0, value)) %>% 
pivot_wider(id_cols = "date",
names_from = "datatype",
values_from = "value") %>% 
mutate(TAVG=(TMIN+TMAX)/2,
month=parse_number(substr(date, 5,6)),
year=parse_number(substr(date, 1,4)),
date=as.Date(date, "%Y%m%d")) %>% 
group_by(month, year) %>% 
summarize(AVG_TAVG=mean(TAVG, na.rm = TRUE),
AVG_PRCP=mean(PRCP, na.rm=TRUE),
AVG_SNOW=mean(SNOW, na.rm=TRUE)) %>% 
ungroup()
#> # A tibble: 7 x 5
#>   month  year AVG_TAVG AVG_PRCP AVG_SNOW
#>   <dbl> <dbl>    <dbl>    <dbl>    <dbl>
#> 1     1  2020    -1.82     6.61   2.58  
#> 2     2  2020    -7.60     9.31  11.3   
#> 3     3  2020    31.6      1.77   0.0968
#> 4     4  2020    69.9     15.1    3.97  
#> 5     5  2020   119.      21.3    0     
#> 6     6  2020   155.      21.5    0     
#> 7     7  2020   191.       2.55   0