我有2个数据框(DF),每个数据框都包含标识符和日期范围。在两个DF中,每个ID可能有很多日期范围。
我想做的是从第二个DF(DF.B)中的第一个DF(DF.A)中选择有任意长度重叠间隔的行。
df.A <- data.frame("ID" = c(1,1,1,2,3,3),
"Start.A" = c("2019-01-01", "2019-03-15", "2019-06-10", "2017-01-01", "2015-05-10", "2015-05-15"),
"End.A" = c("2019-01-31", "2019-04-15", "2019-07-09", "2017-01-31", "2015-06-10", "2015-06-02"))
df.B <- data.frame("ID" = c(1,1,1,3,3),
"Start.B" = c("2019-01-01", "2019-02-01", "2019-03-01", "2015-06-01", "2015-07-01"),
"End.B" = c("2019-01-31", "2019-02-28", "2019-03-31", "2015-06-30", "2015-07-31"))
数据框A:
ID Start.A End.A
1 2019-01-01 2019-01-31
1 2019-03-15 2019-04-15
1 2019-06-10 2019-07-09
2 2017-01-01 2017-01-31
3 2015-05-10 2015-06-10
3 2015-05-15 2015-06-02
数据框B:
ID Start.B End.B
1 2019-01-01 2019-01-31
1 2019-02-01 2019-02-28
1 2019-03-01 2019-03-31
3 2015-06-01 2015-06-30
3 2015-07-01 2015-07-31
我想要我的输出为:
ID Start.A End.A
1 2019-01-01 2019-01-31
1 2019-03-15 2019-04-15
3 2015-05-10 2015-06-10
3 2015-05-15 2015-06-02
我认为,如果我进行一对一比赛,我将能够毫无问题地做到这一点,但是,正如我提到的,在两个DF中,每个ID都有很多观察结果。我已经尝试过尝试使用lubridate的间隔,但是我在如何寻找重叠的同时却苦苦挣扎,同时还要处理因查找潜在匹配而必须在DF.B中查找所有对应ID的复杂性。
这是一个非常大的数据集(DF.A中有500万个观测值,DF.B中有200万个观测值),因此速度至关重要。任何对数据进行转换以使其尽可能快地进行操作的建议也将不胜感激。
如果有帮助:对于给定的ID,DF.A的观察值可以与DF.A中的其他观察值重叠(例如,上面玩具示例中的ID 3)。相反,DF.B间隔之间没有重叠。
答案 0 :(得分:0)
如何?
static function get_video_attributes(&$video_attribute, $video, $ffprobe="ffprobe")
{
$command = "$ffprobe $video 2>&1";
exec($command, $output, $return_var);
if ($return_var) {
Log::channel('queuelog')->error("ERR EXEC: " . $command);
return ERR_FILE;
}
$output = implode("\n",$output);
$regex_sizes = "/Video: ([^,]*), (.*), ([0-9]{1,4})x([0-9]{1,4}).*, ([0-9]{1,4}) kb\/s, (.*) fps, (.*) tbr, (.*) tbn/";
if (preg_match($regex_sizes, $output, $regs)) {
$v_codec = $regs[1] ? $regs[1] : null;
$width = $regs[3] ? $regs[3] : null;
$height = $regs[4] ? $regs[4] : null;
$v_bps = $regs[5] ? $regs[5] : null;
$tbn = $regs[8] ? $regs[8] : null;
}
$regex_sizes = "/Audio: ([^,]*), ([^,]*) Hz, (.*), ([0-9]{1,4}) kb\/s/";
if (preg_match($regex_sizes, $output, $regs)) {
$a_codec = $regs[1] ? $regs[1] : null;
$hz = $regs[2] ? $regs[2] : null;
$a_bps = $regs[4] ? $regs[4] : null;
}
$regex_duration = "/Duration: ([0-9]{1,2}):([0-9]{1,2}):([0-9]{1,2}).([0-9]{1,2}), (.*) ([0-9]{1,4}) kb\/s/";
if (preg_match($regex_duration, $output, $regs)) {
$hours = $regs[1] ? $regs[1] : null;
$mins = $regs[2] ? $regs[2] : null;
$secs = $regs[3] ? $regs[3] : null;
$ms = $regs[4] ? $regs[4] : null;
$bps = $regs[6] ? $regs[6] : null;
}
$video_attribute = array(
'width' => $width ? $width : null,
'height' => $height ? $height : null,
'duration' => ($duration = $hours * 3600 + $mins * 60 + $secs) ? $duration : null,
'hours' => $hours ? $hours : null,
'mins' => $mins ? $mins : null,
'secs' => $secs ? $secs : null,
'ms' => $ms ? $ms : null,
'hz' => $hz ? $hz : null,
'v_codec' => $v_codec ? $v_codec : null,
'a_codec' => $a_codec ? $a_codec : null,
'bps' => $bps ? $bps : null,
'v_bps' => $v_bps ? $v_bps : null,
'a_bps' => $a_bps ? $a_bps : null,
'tbn' => $tbn ? $tbn : null
);
return SC_SUCCESS;
}
还有
library(data.table)
df.A <- data.table("ID" = c(1,1,1,2,3,3),
"Start.A" = c("2019-01-01", "2019-03-15", "2019-06-10", "2017-01-01", "2015-05-10", "2015-05-15"),
"End.A" = c("2019-01-31", "2019-04-15", "2019-07-09", "2017-01-31", "2015-06-10", "2015-06-02"))
df.B <- data.table("ID" = c(1,1,1,3,3),
"Start.B" = c("2019-01-01", "2019-02-01", "2019-03-01", "2015-06-01", "2015-07-01"),
"End.B" = c("2019-01-31", "2019-02-28", "2019-03-31", "2015-06-30", "2015-07-31"))
结果显示如下:
DF = merge(df.A, df.B , by ='ID',allow.cartesian = TRUE)
DF$SEQ_DATE.A = apply(DF[,c('Start.A','End.A'), with=F],1, function(x){paste(x,collapse = ',')})
DF$SEQ_DATE.A = unlist(lapply(strsplit(DF$SEQ_DATE.A,','),function(x){
out = seq(as.Date(x[1]),as.Date(x[2]),by = 'day')
out = paste(out, collapse = '|')
return(out)
}
))
DF$SEQ_DATE.B = apply(DF[,c('Start.B','End.B'), with=F],1, function(x){paste(x,collapse = ',')})
DF$SEQ_DATE.B = unlist(lapply(strsplit(DF$SEQ_DATE.B,','),function(x){
out = seq(as.Date(x[1]),as.Date(x[2]),by = 'day')
out = paste(out, collapse = '|')
return(out)
}
))
DF$Result= apply(DF[,c('SEQ_DATE.A','SEQ_DATE.B'), with = F], 1, function(x){grepl(x[1],x[2])})