1

「日付」列と「データ」列のワイドフォーマットできちんと整理されていない月次時系列データを読み取る方法を研究しています。たとえば、SEMIのこのスプレッドシートには、月と地域ごとに編成されたデータのブロックがありますが、年は区切られ、連続していないブロックにあり、YYYY 形式の年が各ブロックの前のヘッダーとして表示されます。

不連続データ

私の目的は、このデータを、列 1 に月次日付、列 2:6 に地域データを含む連続したブロックに変換することです。このスプレッドシートをタブ区切りファイルとしてエクスポートした後 (両方ともgdata、XLConnectスクリーンショットに見られるような結合されたセルに問題があることがわかりました)、それを読み込んで、dput以下のソースであるサブセットを取得しました。

次のようなものを使用して、最初に空の行を取り除くというアプローチを取りました。

mydf <- mydf[which(grepl("^$", mydf$January) == FALSE),]

次に、年を含む行の [地域] 列にラベルを追加します。便利なことに、これは常に 2 番目 ('1 月') の列に表示されます。

mydf[which(nchar(mydf$January) == 4) ,'Region'] <- 'mydate'

次のステップは、これらの「年」行の 1 月から 12 月までの列に月の日付を入力することです。毎月固有の日付があればddply、何かを使用して処理できると思いました。

mydf[which(mydf$Region == 'mydate'), 2:13] <- apply(mydf[which(mydf$Region == 'mydate'), 2:13], 1, function(x) as.character(seq(as.Date(paste(x['January'],"-01-01", sep = "")), as.Date(paste(x['January'],"-12-01", sep = "")), by = 'month')))

関数が私が望んでいた方法で日付を生成していないため、これは期待どおりにapply機能していません-それらは順番にありません。apply（a）ステップの特定の修正、または（b）より単純または簡単なポインターの代替アプローチのいずれかをいただければ幸いです。

以下のデータとコード:

mydf <- structure(list(Region = c("", "Americas", "Europe", "Japan",
"Asia Pacific", "Worldwide", "", "", "Americas", "Europe", "Japan",
"Asia Pacific", "Worldwide", "", "", "Americas", "Europe", "Japan",
"Asia Pacific", "Worldwide", "", "", "", "Americas", "Europe",
"Japan", "Asia Pacific", "Worldwide", "", "", "Americas", "Europe",
"Japan", "Asia Pacific", "Worldwide"), January = c("1980", "413136",
"189577", "34033", "39868", "676614", "", "1981", "445504", "277290",
"33970", "44642", "801406", "", "1982", "445300", "226274", "34404",
"44989", "750967", "", "January", "1983", "457604", "232443",
"34326", "46247", "770621", "", "1984", "731009", "285740", "205644",
"85426", "1307820"), February = c("", "423748", "234818", "35104",
"42398", "736069", "", "", "440225", "274526", "33795", "44005",
"792550", "", "", "438332", "226806", "33359", "44020", "742517",
"", "February", "", "457899", "233560", "32604", "46184", "770247",
"", "", "790963", "307735", "381282", "102791", "1582770"), March = c("",
"436152", "281353", "34456", "46555", "798516", "", "", "434628",
"267259", "33709", "45206", "780802", "", "", "441313", "235612",
"32380", "43600", "752905", "", "March", "", "459498", "234986",
"31544", "48178", "774206", "", "", "856970", "339674", "574527",
"118091", "1889262"), April = c("", "455673", "288710", "34451",
"48585", "827419", "", "", "443285", "264405", "34823", "47192",
"789705", "", "", "465613", "246425", "33618", "46274", "791930",
"", "April", "", "484299", "243867", "32719", "52333", "813218",
"", "", "909873", "364465", "627400", "126954", "2028693"), May = c("",
"474441", "297343", "35092", "51102", "857977", "", "", "451221",
"255887", "35499", "48459", "791065", "", "", "487738", "249522",
"34339", "47727", "819325", "", "May", "", "507807", "246136",
"34708", "59300", "847950", "", "", "969553", "382706", "655862",
"133455", "2141576"), June = c("", "475552", "299427", "35743",
"51440", "862162", "", "", "453152", "242889", "35798", "48147",
"779986", "", "", "488564", "241273", "34360", "48871", "813068",
"", "June", "", "528620", "246710", "37345", "62910", "875586",
"", "", "991274", "388697", "672773", "135550", "2188294"), July = c("",
"473007", "302075", "37771", "51027", "863880", "", "", "454387",
"231097", "35402", "47468", "768353", "", "", "480702", "229555",
"33915", "49112", "793284", "", "July", "", "543063", "241211",
"40403", "66658", "891335", "", "", "1005742", "395852", "683854",
"138853", "2224302"), August = c("", "462125", "294497", "37628",
"49773", "844023", "", "", "450648", "213017", "34363", "46614",
"744642", "", "", "472486", "215763", "32866", "48620", "769734",
"", "August", "", "565034", "236353", "42524", "66853", "910763",
"", "", "1010739", "393337", "691731", "141101", "2236908"),
    September = c("", "461968", "295501", "37310", "50280", "845059",
    "", "", "459276", "215403", "33801", "47297", "755777", "",
    "", "475729", "219643", "33083", "47540", "775994", "", "September",
    "", "593019", "244979", "44108", "70242", "952348", "", "",
    "1035725", "408658", "698992", "141944", "2285320"), October = c("",
    "459862", "296522", "36399", "51220", "844003", "", "", "465096",
    "218792", "34168", "47369", "765424", "", "", "467151", "225828",
    "33667", "47890", "774536", "", "October", "", "618854",
    "259807", "47622", "71345", "997628", "", "", "1033560",
    "421043", "710563", "140154", "2305320"), November = c("",
    "456832", "296283", "35769", "50531", "839415", "", "", "467288",
    "232593", "35039", "47415", "782335", "", "", "461950", "237117",
    "35672", "47285", "782024", "", "November", "", "641864",
    "275099", "50371", "72095", "1039428", "", "", "1008836",
    "441652", "732948", "133861", "2317297"), December = c("",
    "460343", "291348", "35781", "48298", "835771", "", "", "460574",
    "231461", "35971", "47173", "775179", "", "", "462919", "235861",
    "36251", "47974", "783006", "", "December", "", "672533",
    "276525", "54603", "74717", "1078379", "", "", "982210",
    "442448", "731546", "132982", "2289187")), .Names = c("Region",
"January", "February", "March", "April", "May", "June", "July",
"August", "September", "October", "November", "December"), row.names = 29:63, class = "data.frame")

mydf <- mydf[which(grepl("^$", mydf$January) == FALSE),] # remove rows with nothing in the January column
mydf[which(nchar(mydf$January) == 4) ,'Region'] <- 'mydate' # add a row label for 'year' rows

mydf[which(mydf$Region == 'mydate'), 2:13] <- apply(mydf[which(mydf$Region == 'mydate'), 2:13], 1, function(x) as.character(seq(as.Date(paste(x['January'],"-01-01", sep = "")), as.Date(paste(x['January'],"-12-01", sep = "")), by = 'month')))
4

3 に答える 3

3

あなたが使用することができxlsReadWriteますreshape2

 library(xlsReadWrite)
 tdata<-read.xls('GSR1976-June 2012.xls',stringsAsFactors=F)
 tdata[85,2]<-1987 # fix for missing year
 tdata[228,2]<-2007 # fix for missing year
 year.marker<-c(grep('^[[:digit:]]{4}$',tdata[,2]),270)

 temp.df<-NULL

 for(i in seq_along(year.marker)[-length(year.marker)]){
   dum.df<-cbind(tdata[year.marker[i],2],tdata[(year.marker[i]+1):(year.marker[i+1]-2),])
   temp.df<-rbind(temp.df,dum.df)
 }

 names(temp.df)<-c('year','region',month.name)

 df1<-temp.df[!temp.df[,'region']=='',]
 library(reshape2)
 df2<-melt(df1, id.vars=c("region", "year"))
于 2012-08-18T12:35:27.640 に答える
2

私は次のアプローチを取りました。

まず、あなたのファイルを CSV に変換し、次に行を読み込みました。grep()以前は、各セットの最初の行である「Americas」を見つけていました。開始年と終了年を手動で入力しましたが、grepおそらくそこでも使用できるものもあります。

temp = readLines("GSR1976-June 2012.csv")
START = grep("Americas", temp)
YEARS = 1976:2012

その後、data.frame各年に 1 つずつ、s のリストを作成しました。

temp1 = lapply(1:length(YEARS), 
               function(x) read.csv("GSR1976-June 2012.csv",
                                    header=FALSE, skip=START[x]-1,
                                    nrows=5))
names(temp1) = YEARS

次に、それらを 1 つにまとめて、data.frameクリーンアップを行いました。

temp2 = do.call(rbind, temp1)
names(temp2) = c("region", "jan", "feb", "mar", "apr", "may", "jun",
                 "jul", "aug", "sep", "oct", "nov", "dec")
temp2$year = rep(YEARS, each=5)

やりたい再形成のタイプは指定しませんが、ワイドからロングに変更したい場合は、reshape2パッケージを使用するのが最も簡単な方法です。

library(reshape2)
temp3 = melt(temp2, id.vars=c("region", "year"))
list(head(temp3), tail(temp3))
# [[1]]
#         region year variable  value
# 1     Americas 1976      jan     NA
# 2       Europe 1976      jan     NA
# 3        Japan 1976      jan     NA
# 4 Asia Pacific 1976      jan     NA
# 5    Worldwide 1976      jan     NA
# 6     Americas 1977      jan 195638
# 
# [[2]]
#            region year variable    value
# 2215    Worldwide 2011      dec 23832532
# 2216     Americas 2012      dec       NA
# 2217       Europe 2012      dec       NA
# 2218        Japan 2012      dec       NA
# 2219 Asia Pacific 2012      dec       NA
# 2220    Worldwide 2012      dec       NA

次に、探しているように聞こえるdcast()出力については、次を使用します。

temp4 = dcast(temp3, year + variable ~ region)
head(temp4)
#   year variable Americas Asia Pacific Europe Japan Worldwide
# 1 1976      jan       NA           NA     NA    NA        NA
# 2 1976      feb       NA           NA     NA    NA        NA
# 3 1976      mar   178295        16761  55602 10805    261463
# 4 1976      apr   178961        16513  60959 11589    268022
# 5 1976      may   187076        17396  62329 12435    279235
# 6 1976      jun   193675        17712  61676 14411    287475
于 2012-08-18T12:28:32.217 に答える
1

上記のデータ セットは、次のようにXLConnectを使用して Excel ファイルから直接簡単に処理できます。

require(XLConnect)
require(reshape2)

# Load Excel workbook
wb = loadWorkbook("~/Downloads/GSR1976-June 2012.xls")

# Read data from 1st worksheet, starting at row 7 with predefined column types
data = readWorksheet(wb, sheet = 1, startRow = 7, 
    colTypes = c("character", rep("numeric", 12)))
# Rename first column and keep month names
colnames(data)[1] = "Region"
months = names(data)[-1]

# The data of merged cells (years) is in the first cell of the merged region
years = ifelse(is.na(data$Region), data$January, NA)
idx = !is.na(years)

# Replicate year information to form a new column 'Year'
data$Year = rep(years[idx], times = diff(c(which(idx), length(years) + 1)))

# Remove any rows where 'Region' is missing (^= non-data rows)
data = data[!is.na(data$Region), ]

# Reshape (wide --> long)
data = melt(data, measure.vars = months, variable.name = "Month")
于 2012-08-19T19:18:04.173 に答える