绘图覆盖，所以在 R 循环结束时，都从列表的最后一个元素中拉出。我究竟做错了啥？
Posted
技术标签:
【中文标题】绘图覆盖，所以在 R 循环结束时，都从列表的最后一个元素中拉出。我究竟做错了啥？【英文标题】：Plots overwriting, so at the end of the R loop all have pulled from last element of list. What am I doing wrong?绘图覆盖，所以在 R 循环结束时，都从列表的最后一个元素中拉出。我究竟做错了什么？ 【发布时间】：2021-01-08 14:09:34 【问题描述】：我已经在这个循环上停留了一段时间（从我的问题历史可以看出），但我想我已经接近修复它了，非常感谢我在堆栈溢出方面获得的帮助。
我注意到在我的绘图中，每个绘图都使用 data_percentage_list[391]，即列表中的最后一个元素。我做了很多事情来试图阻止这种情况的发生，但是使用下面的代码：
# Create graphs in list
# Create titles for plots
titlenames <- c(harps)
for (i in 1:length(harps))
counts <- table(Y[[i]][[5]], Y[[i]][[3]])
nam <- paste("data_percentage_", i, sep ="")
assign(nam, apply(counts, 2, function(x)x*100/sum(x,na.rm=T)))
data_percentage_list <- lapply(paste0("data_percentage_",1:length(harps)), get)
# Create pdf of score breakdown
for (i in 1:length(harps)) for(j in titlenames)
# For Hotel Name Subtitle
hotelname <- hotel_report$`Hotel (Q15 1)`[hotel_report$`Harp Number`==j]
# Plot the Data
pdf(file = paste0(j, ".pdf"), paper = "USr", width=8, height=7)
par(mar = c(5.1, 7, 4.1, 2.1))
nam <- paste("breakdown_", i, sep ="")
assign(nam, barplot(data_percentage_list[[i]], main = "Breakdown of Property Score Distribution", sub = hotelname,
col = coul, las = 1, cex.names = .6, horiz = TRUE, yaxs="i", xlab = "Percentage",
cex.axis = .8, cex.lab = .8, cex.main = .8, cex.sub = .8))
dev.off()
其中 length(harps) 为 391，因此有 391 个图，这些图在进行时会被覆盖。因此，当我在 pdf 中打开一个绘图并刷新它时，它会更改为循环的最后一次迭代，直到最后它们都成为第 391 个属性的数据，并且具有正确的酒店名称，因为它是从 j 中提取的。
有谁知道我需要如何更改我的代码以使每个图都对应于正确的数据？意思是，breakdown_54 应该使用 data_percentage_list[54]，并保存为该数据的 pdf，breakdown_55 应该是 data_percentage_list[55]，等等？
谢谢！
编辑：在进一步处理之后跟进。
下面的代码生成了 391 个不同的图表，但 391 个 pdf 文件中的每一个都包含所有 391 个图表，而不是像他们应该的那样只有各自的图表。
在此代码中正确拆分这些 pdf 是否比修复上面的代码更容易？
# Create graphs in list
# Create titles for plots
titlenames <- c(harps)
for (i in 1:length(harps))
counts <- table(Y[[i]][[5]], Y[[i]][[3]])
nam <- paste("data_percentage_", i, sep ="")
assign(nam, apply(counts, 2, function(x)x*100/sum(x,na.rm=T)))
data_percentage_list <- lapply(paste0("data_percentage_",1:length(harps)), get)
# Create pdf of score breakdown
for (i in 1:length(harps)) for(j in titlenames)
# For Hotel Name Subtitle
hotelname <- hotel_report$`Hotel (Q15 1)`[hotel_report$`Harp Number`==j]
# Plot the Data
pdf(file = paste0(j, ".pdf"), paper = "USr", width=8, height=7)
par(mar = c(5.1, 7, 4.1, 2.1))
nam <- paste("breakdown_", i, sep ="")
breakdown_list <- lapply(1:length(harps), function(i)
assign(nam, barplot(data_percentage_list[[i]], main = "Breakdown of Property Score Distribution", sub = hotelname,
col = coul, las = 1, cex.names = .6, horiz = TRUE, yaxs="i", xlab = "Percentage",
cex.axis = .8, cex.lab = .8, cex.main = .8, cex.sub = .8)))
dev.off()
再次感谢！
编辑 2：试图使其更具重现性
Y 是 391 个数据帧的列表 下面的代码是 Y 中 391 个数据帧之一的 dput。
structure(list(`Hotel (Q15 1)` = c("HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE", "HILTON, SAN PEDRO, BELIZE",
"HILTON, SAN PEDRO, BELIZE"), `Metro Area State (Q10 1)` = c("OCONUS",
"OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS",
"OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS",
"OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS", "OCONUS",
"OCONUS"), `Question ID` = c("Room Work Area", "Staff Knowledge",
"Add'tl Item Working Order", "Property Maintenance", "Property Appearance",
"Staff Knowledge", "Property Appearance", "Staff Interaction",
"Safety/Security", "Add'tl Item Working Order", "Room Work Area",
"Bed Quality", "Check In/Out", "Invoice Accuracy", "Staff Interaction",
"Safety/Security", "Bed Quality", "Invoice Accuracy", "Check In/Out",
"Safety/Security", "Invoice Accuracy", "Bed Quality", "Property Maintenance"
), `Question ID (group)` = c("Question 4 Items", "Question 4 Items",
"Question 4 Items", "Question 4 Items", "Question 4 Items", "Question 4 Items",
"Question 4 Items", "Question 4 Items", "Question 4 Items", "Question 4 Items",
"Question 4 Items", "Question 4 Items", "Question 4 Items", "Question 4 Items",
"Question 4 Items", "Question 4 Items", "Question 4 Items", "Question 4 Items",
"Question 4 Items", "Question 4 Items", "Question 4 Items", "Question 4 Items",
"Question 4 Items"), `Score Label` = c("7 Extremely Good", "7 Extremely Good",
"7 Extremely Good", "7 Extremely Good", "7 Extremely Good", "6 Quite Good",
"6 Quite Good", "6 Quite Good", "6 Quite Good", "6 Quite Good",
"6 Quite Good", "6 Quite Good", "6 Quite Good", "7 Extremely Good",
"7 Extremely Good", "5 Slightly Good", "7 Extremely Good", "6 Quite Good",
"7 Extremely Good", "7 Extremely Good", "3 Slightly Poor", "5 Slightly Good",
"6 Quite Good"), `Harp Number` = c("1111", "1111", "1111", "1111",
"1111", "1111", "1111", "1111", "1111", "1111", "1111", "1111",
"1111", "1111", "1111", "1111", "1111", "1111", "1111", "1111",
"1111", "1111", "1111")), row.names = c(9380L, 9381L, 9383L,
9384L, 9385L, 9387L, 9388L, 9389L, 9390L, 9391L, 9392L, 9393L,
9394L, 9395L, 9396L, 9399L, 9402L, 9403L, 9404L, 9405L, 9407L,
9408L, 9411L), class = "data.frame")
下面是 dput(harps)
dput(harps)
c("1111", "1696", "3279", "5646", "5724", "5938", "6887", "8859",
"9368", "9508", "11569", "11644", "18661", "21418", "22460",
"23317", "25755", "26076", "26336", "28917", "29497", "29498",
"30465", "30619", "30629", "32784", "35578", "35588", "40390",
"40866", "47493", "47677", "47866", "48064", "48294", "50432",
"50667", "50773", "51857", "52125", "52146", "52383", "52432",
"52451", "52755", "53589", "53620", "56939", "57784", "59571",
"61276", "61283", "62329", "62666", "66058", "66553", "66741",
"66763", "67092", "67169", "67214", "67373", "67840", "69494",
"71343", "73906", "74550", "75285", "76253", "76335", "76361",
"76393", "76396", "76898", "76949", "78501", "78800", "80079",
"81035", "81620", "85043", "87026", "87219", "87304", "88683",
"89650", "92759", "94380", "94427", "95043", "95255", "96061",
"96677", "97269", "100135", "109591", "109743", "109971", "110414",
"110856", "110884", "110899", "110926", "111032", "111384", "111605",
"123136", "123411", "124380", "124753", "124848", "127565", "135185",
"135999", "136005", "138251", "140027", "140074", "140091", "140095",
"140159", "145523", "148284", "149639", "153676", "154790", "157239",
"158213", "158259", "159248", "159343", "159401", "159842", "161219",
"161725", "163154", "163653", "167172", "170199", "171936", "172095",
"172272", "172273", "172340", "172868", "173429", "173816", "175033",
"177012", "177150", "177361", "177383", "177692", "177892", "177965",
"179887", "180495", "182189", "182979", "183174", "183717", "183879",
"184076", "185191", "185341", "185675", "185961", "189276", "190279",
"190896", "192388", "192984", "193387", "193441", "193526", "193534",
"193605", "193613", "193614", "194274", "194794", "196133", "196546",
"197075", "197647", "198115", "200996", "201627", "202124", "202992",
"205802", "206405", "206880", "206990", "207423", "207483", "207723",
"208210", "208943", "209614", "210006", "211605", "211985", "212714",
"213707", "213803", "213842", "215961", "216533", "217963", "218029",
"218348", "218376", "221745", "222179", "222299", "222399", "222736",
"222882", "224539", "224624", "225339", "225346", "225368", "225553",
"225565", "225572", "225573", "226003", "228325", "229582", "229614",
"230871", "231228", "231402", "235196", "235538", "239409", "241353",
"244587", "244654", "245353", "246093", "246311", "247209", "251084",
"253732", "254388", "256996", "258464", "260958", "261655", "262754",
"263192", "263444", "265835", "269872", "270285", "271683", "271687",
"272664", "275922", "276312", "279909", "287731", "291167", "291988",
"296004", "297975", "298318", "298401", "300962", "301940", "302250",
"302702", "304896", "308049", "311490", "312027", "313227", "313603",
"315536", "319957", "320049", "320270", "320352", "327521", "330319",
"331054", "332070", "332426", "334213", "341876", "345820", "346263",
"346723", "347340", "352596", "354486", "396465", "445549", "473263",
"482701", "496665", "503123", "503365", "528259", "538396", "539834",
"540896", "546228", "546290", "546652", "546922", "548916", "550479",
"552466", "709416", "714793", "714861", "716337", "719021", "728913",
"731082", "732346", "733242", "735165", "735348", "735473", "749296",
"757777", "761782", "762104", "770251", "808540", "809896", "809951",
"812527", "816275", "837926", "842678", "843836", "847737", "857277",
"864044", "864495", "865468", "865951", "866108", "866502", "866547",
"867803", "867809", "868374", "868420", "868593", "868793", "869746",
"869748", "870953", "872490", "872579", "875200", "875288", "878016",
"878858", "879328", "879640", "882643", "882781", "883894", "886067",
"886876", "888522", "888560", "888820", "889693", "890261", "890264",
"891171", "894931", "896794", "896840", "899485", "901218", "903465",
"904381", "912517", "913354", "918968", "921083")
【问题讨论】：
我认为问题在于您的 pdf 部分中有 2 个 for 循环（参见“for (i in 1:length(harps)) for(j in titlenames) ” 代码行）。因此，对于每个竖琴 i，您都在遍历整个标题集。然后，每个竖琴的 pdf 每次都与最后一个标题相关联。如果你把它减少到一个循环，它会更好地为你工作。 嗨@olorcain，谢谢！我试图减少到一个循环，但它给了我太多打开的设备错误，我无法修复。 在您最近没有官方答案的问题历史中，除了我们无法轻松导入的small screenshot 之外，您没有提供示例数据。请提供Y 的dput 并显示harps。所有发布的代码都应该可以从空的 R 环境中运行。这里的一些编码行可以在不使用assign 和get 以及多个for 循环的情况下收紧。见How to make a great R reproducible example。我们热爱数据并会为您解答！
@Parfait，谢谢！我以前从未听说过 dput 。再次编辑，希望它更具重现性。由于隐私问题，无法将所有 Y 的 dput 放在那里，但将 Y 元素之一的编辑版本的 dput 放在那里。
【参考方案1】：
考虑以下 R 语言甚至编程的一般技巧：
变量：避免使用太多变量，而是直接与现有对象交互。这增强了环境变量的可维护性。一些冗余示例包括：
titlenames <- c(harps)
nam <- paste("data_percentage_", i, sep ="")
data_percentage_list <- lapply(paste0("data_percentage_",1:length(harps)), get)
名称：为对象使用更具信息性的名称，因为 Y 不会在未来通知代码读者或您自己。它似乎是一个列表，其中包含较大数据框hotel_report 的子集。 hotel_reports_df_list 等信息更丰富的名称可以快速详细说明其内容和类型（即列表中的数据框）。
缩进：总是在 for 循环中缩进代码（可以在 RStudio 中使用键自动缩进：Ctrl/cmd + i) 甚至在 pdf、with 等上下文管理器内部。这增强了可读性和可维护性。
分配/获取：避免在 R 中通常不推荐使用的 assign 和 get。相反，直接将对象保存为列表中的项目。第一个循环可以绕过将子项分配为单独变量的需要：
data_pct_matrix_list <- lapply(seq_along(harps), function(i)
counts <- table(Y[[i]][[5]], Y[[i]][[3]])
pct_matrix <- apply(counts, 2, function(x) x*100/sum(x, na.rm=TRUE))
return(pct_matrix)
)
最后assign包裹barplot也可以重构：
plot_list <- lapply(data_percentage_matrix_list, function(mat)
barplot(mat, main = "Breakdown of Property Score Distribution", sub = hotelname,
col = coul, las = 1, cex.names = .6, horiz = TRUE, yaxs="i", xlab = "Percentage",
cex.axis = .8, cex.lab = .8, cex.main = .8, cex.sub = .8))
)
循环：尽可能避免多个for 或嵌套循环。在 R 中，lapply 是一个隐藏循环。您在 391 个 PDF 中的 391 个绘图的问题可能是由于嵌套的 lapply within for 循环。考虑以下步骤：
-
首先，考虑您对一个数据框对象的处理。甚至可以将其概括为一个单独的函数。
然后，想想究竟哪些变化可以迭代。
R 的 apply 系列不仅包括 apply 和 lapply，例如 mapply 可以运行元素循环以展平嵌套迭代或 by（面向对象的包装器到 tapply）可以子集数据按因子列划分帧并对它们运行操作。
在不查看示例数据的情况下，考虑以下需要针对数据进行测试的方法。下面假设 Y 被定义为来自 hotel_report 数据帧的子集列表 Harp Number。
mapply / Map 方法
在等长对象data_pct_matrix_list和harps之间逐元素迭代。
data_pct_matrix_list <- lapply(seq_along(harps), function(i)
counts <- table(Y[[i]]$`Score Label`, Y[[i]]$`Question ID`)
pct_matrix <- apply(counts, 2, function(x) x*100/sum(x, na.rm=TRUE) )
return(pct_matrix)
)
build_pdf <- function(data, harp)
# For Hotel Name Subtitle
hotelname <- hotel_report$`Hotel (Q15 1)`[hotel_report$`Harp Number` == harp]
# Plot the Data
pdf(file = paste0(harp, ".pdf"), paper = "USr", width=8, height=7)
par(mar = c(5.1, 7, 4.1, 2.1))
hotel_plot <- barplot(data, main = "Breakdown of Property Score Distribution", sub = hotelname,
col = coul, las = 1, cex.names = .6, horiz = TRUE, yaxs="i", xlab = "Percentage",
cex.axis = .8, cex.lab = .8, cex.main = .8, cex.sub = .8))
dev.off()
return(hotel_plot)
plot_list <- Map(build_pdf, data_pct_matrix_list, harps)
# EQUIVALENTLY:
plot_list <- mapply(build_pdf, data_pct_matrix_list, harps, SIMPLIFY=FALSE)
by 方法
子集 hotel_report 数据帧由唯一的 Harp Number 组成，并在每个子集上迭代运行以构建 pct_matrix 和 hotel_plot。这种方法结合了矩阵构建和绘图步骤。
build_pdf <- function(sub_df)
# Matrix build
counts <- table(sub_df$`Score Label`, sub_df$`Question ID`)
pct_matrix <- apply(counts, 2, function(x) x*100/sum(x, na.rm=TRUE) )
# For Hotel Name Subtitle
hotelname <- sub_df$`Hotel (Q15 1)`[1]
harp <- sub_df$`Harp Number`[1]
# Plot the Data
pdf(file = paste0(harp, ".pdf"), paper = "USr", width=8, height=7)
par(mar = c(5.1, 7, 4.1, 2.1))
hotel_plot <- barplot(pct_matrix, main = "Breakdown of Property Score Distribution", sub = hotelname,
col = coul, las = 1, cex.names = .6, horiz = TRUE, yaxs="i", xlab = "Percentage",
cex.axis = .8, cex.lab = .8, cex.main = .8, cex.sub = .8))
dev.off()
return(hotel_plot)
plot_list <- by(hotel_report, hotel_report$`Harp Number`, build_pdf)
# NEAR EQUIVALENT
plot_list <- lapply(split(hotel_report, hotel_report$`Harp Number`), build_pdf)
【讨论】：
非常感谢！这种 mapply 方法非常有效！我永远不会想到这一点。并感谢您提供的所有编程技巧！我肯定会重新检查我的代码并修复它，因为它完全是一团糟。以上是关于绘图覆盖，所以在 R 循环结束时，都从列表的最后一个元素中拉出。我究竟做错了啥？的主要内容，如果未能解决你的问题，请参考以下文章