查看本文末尾的更新
我正在使用R来分析来自小鼠实验的车轮运行数据。我有以下数据框：
daily diff diff_abs RSI revRSI date
31115 0 0 NA 41.2414625757225 01/25
26455 -4660 4660 NA 48.2658030703815 01/26
43746 17291 17291 NA 45.9494803496455 01/27
56477 12731 12731 NA 48.9645147955033 01/28
20487 -35990 35990 NA 50.8358888804281 01/29
48091 27604 27604 NA 27.3308517626451 01/30
40780 -7311 7311 NA 55.0797465588814 01/31
66862 26082 26082 NA 47.1060765697854 02/01
56070 -10792 10792 58.7585374242775 54.7090598105694 02/02
36727 -19343 19343 51.7341969296185 52.8298034135076 02/03
39402 2675 2675 54.0505196503545 36.9980600069096 02/04
46852 7450 7450 51.0354852044967 50.6400779877448 02/05
54062 7210 7210 49.1641111195719 48.523901908252 02/06
82605 28543 28543 72.6691482373549 48.3957328746788 02/07
31816 -50789 50789 44.9202534411186 53.2082055322621 02/08
50723 18907 18907 52.8939234302146 34.3408339939252 02/09
52931 2208 2208 45.2909401894306 49.3214157131005 02/10
48032 -4899 4899 47.1701965864924 48.7951434058106 02/11
75867 27835 27835 63.0019399930904 45.400714012296 02/12
37012 -38855 38855 49.3599220122552 52.3590727677325 02/13
52604 15592 15592 51.476098091748 33.3383758162519 02/14
60330 7726 7726 51.6042671253212 52.719803210509 02/15
71204 10874 10874 46.7917944677379 58.8525137798564 02/16
71719 515 515 65.6591660060748 71.5165371132449 02/17
52457 -19262 19262 50.6785842868995 76.8903074849309 02/18
56043 3586 3586 51.2048565941894 64.6605567935721 02/19
59807 3764 3764 54.599285987704 71.2648737251093 02/20
70630 10823 10823 47.6409272322675 73.3683338802169 02/21
63446 -7184 7184 66.6616241837481 89.0878793194177 02/22
48314 -15132 15132 47.280196789491 61.6081268190681 02/23
47610 -704 704 41.1474862201436 43.4304453935326 02/24
42970 -4640 4640 28.4834628867551 34.2488899851998 02/25
29427 -13543 13543 23.109692515069 27.2125091182907 02/26
33773 4346 4346 35.3394432064279 41.9740001500713 02/27
28021 -5752 5752 28.7351262748907 39.5260512811454 02/28
29896 1875 1875 26.6316661197831 40.432047746235 03/01
26062 -3834 3834 10.9121206805824 38.1126164545335 03/02
47014 20952 20952 38.3918731809319 32.1529563466628 03/03
56928 9914 9914 56.5695546064674 51.6408833585561 03/04
73152 16224 16224 65.7511100148002 55.2920789479607 03/05
81706 8554 8554 72.7874908817093 60.3469873993958 03/06
46541 -35165 35165 58.0259998499287 56.7947848600666 03/07
57490 10949 10949 60.4739487188546 39.5527595903204 03/08
50016 -7474 7474 59.567952253765 55.3361524561763 03/09
58886 8870 8870 61.8873835454665 48.9233030011869 03/10
73397 14511 14511 67.8470436533372 53.8702524564152 03/11
42330 -31067 31067 48.3591166414439 50.062386569873 03/12
42819 489 489 44.7079210520393 31.5655846006373 03/13
47877 5058 5058 39.6530126006042 50.021484430949 03/14
64070 16193 16193 43.2052151399334 54.2381247338469 03/15
66901 2831 2831 60.4472404096796 65.200747697026 03/16
46032 -20869 20869 44.6638475438237 70.6002913025385 03/17
52302 6270 6270 51.0766969988131 58.9211826366076 03/18
51276 -1026 1026 46.1297475435848 61.7373482579221 03/19
73265 21989 21989 49.937613430127 61.4629369889115 03/20
70792 -2473 2473 68.4344153993627 87.5024084356554 03/21
42774 -28018 28018 49.978515569051 78.8086504788302 03/22
39119 -3655 3655 45.7618752661531 80.5670973928131 03/23
36909 -2210 2210 34.799252302974 NA 03/28
27299 -9610 9610 29.3997086974615 NA 03/29
31802 4503 4503 41.0788173633924 NA 03/30
34434 2632 2632 38.2626517420779 NA 03/31
33950 -484 484 38.5370630110885 NA 04/01
30444 -3506 3506 12.4975915643446 NA 04/02
36077 5633 5633 21.1913495211698 NA 04/03
2607 -33470 33470 19.4329026071869 NA 04/04
（NA应该在那里，归因于基于RSI计算revRSI和daily的方式
我有兴趣找到满足以下条件的时间间隔：
RSI < 50连续9次观察或更多RSI < 25进行至少1次观察第一次观察RSI < 50是区间中的第一个数据点。
间隔随观察revRSI > 50而停止（不包括在间隔中）。
如果在>9 RSI < 50 RSI > 50之后revRSI > 50 revRSI > 50之前出现1/0 ，则间隔不会终止。它应该只在TRUE/FALSE时终止。
如果可能的话，我希望输出显示在添加到发布的数据框中的额外列中（类似值为FIRST/LAST或for (n in names(daily)){
animal$daily <- daily[,n]
animal$diff <- ave(animal$daily, FUN=function(x) c(0, diff(x)))
animal$diff_abs <- abs(ave(animal$daily, FUN=function(x) c(0, diff(x))))
animal$diff[animal$diff < 0] <- 0
animal$RSI <- abs(100 * rollsumr(animal$diff, k = 9, fill = NA, align="right") / rollsumr(animal$diff_abs, k = 9, fill = NA, align="right"))
animal$diff <- ave(animal$daily, FUN=function(x) c(0, diff(x)))
animal$diff[animal$diff > 0] <- 0
animal$revRSI <- abs(100 * rollsumr(animal$diff, k = 9, fill = NA, align="left") / rollsumr(animal$diff_abs, k = 9, fill = NA, align="left"))
animal$diff <- ave(animal$daily, FUN=function(x) c(0, diff(x)))
animal$date = datax
animal$indi <- cut(animal$RSI,c(-Inf,50,Inf),c("L50","M50"))
animal$id <- 1:length(animal$RSI)
ind <- which(animal$indi == "M50" | is.na(animal$RSI))
ind <- c(ind,length(animal$RSI) + 1)
final_df <- NULL
for(i in ind){
stop = which(i == ind)
start = ifelse(stop == 1,1,stop-1)
res <- animal[(ind[start] + 1):(ind[stop] -1) ,]
# We create 'little dataframe between each succesiv upper value
if(length(res[,1]) >= 9 ) {
# We flag OK all element
res$FL <- "OK"
# We flag the first element of our little df
res[1,"FL"] <- "FIRST"
# We flag the first element of our little df
res[length(res[,1]),"FL"] <- "LAST"
# Finaly we add our new df to our results
final_df <- rbind(final_df,res[,c("id","FL")])
}
}
animal <- merge(animal,final_df,by.x = "id",by.y="id",all.x =T)
assign(paste("animal", n, sep = ''), animal)
}
或Error in fix.by(by.y, y) : 'by' must specify a uniquely valid column
以指示间隔）。
我尝试搜索，但无法找到与此相关的任何内容。 我希望有人可能有一些好的建议，或者可以指出我正确的方向:) 我将对30只动物进行半年的观察，所以我希望我不必手动这样做;）
谢谢！
编辑在这里开始 我试图将建议合并到我现有的代码中。我正在运行for循环，并希望它成为循环的一部分。 代码（添加新代码）如下所示：
animal我得到的错误是：
FL.x我可以在FL.y data.frame中看到生成FL列，但也有名为-Dlog4j.configuration=/your/config.properties和{{1}}的列。
我不确定那是怎么回事......
另外，我最初忘了提到每只动物没有或多个间隔的可能性（真的应该有，对不起），并且在查看建议的解决方案后，我不确定它是否可以容纳这个......？< / p>
谢谢！
答案 0 :(得分：1)
想法是将您的数据帧与每个RSI值分开，其中RSI>通过这个技巧，您可以评估每个小数据帧的大小。并将df保持在大于9或1的位置。
识别
df$indi <- cut(df$RSI,c(-Inf,50,Inf),c("L50","M50"))
df$id <- 1:length(df$RSI)
ind <- which(df$indi == "M50" | is.na(df$RSI))
ind <- c(ind,length(df$RSI) + 1)
final_df <- NULL
开始执行
for(i in ind){
stop = which(i == ind)
start = ifelse(stop == 1,1,stop-1)
res <- df[(ind[start] + 1):(ind[stop] -1) ,]
# We create 'little dataframe between each succesiv upper value
if(length(res[,1]) >= 9 ) {
# We flag OK all element
res$FL <- "OK"
# We flag the first element of our little df
res[1,"FL"] <- "FIRST"
# We flag the first element of our little df
res[length(res[,1]),"FL"] <- "LAST"
# Finaly we add our new df to our results
final_df <- rbind(final_df,res[,c("id","FL")])
}
}
收集我们的结果
df <- merge(df,final_df,by.x = "id",by.y="id",all.x =T)
结果
id daily diff diff_abs RSI revRSI date indi FL
26 26 56043 3586 3586 51.20486 64.66056 févr-19 M50 <NA>
27 27 59807 3764 3764 54.59929 71.26487 févr-20 M50 <NA>
28 28 70630 10823 10823 47.64093 73.36833 févr-21 L50 <NA>
29 29 63446 -7184 7184 66.66162 89.08788 févr-22 M50 <NA>
30 30 48314 -15132 15132 47.28020 61.60813 févr-23 L50 FIRST
31 31 47610 -704 704 41.14749 43.43045 févr-24 L50 OK
32 32 42970 -4640 4640 28.48346 34.24889 févr-25 L50 OK
33 33 29427 -13543 13543 23.10969 27.21251 févr-26 L50 OK
34 34 33773 4346 4346 35.33944 41.97400 févr-27 L50 OK
35 35 28021 -5752 5752 28.73513 39.52605 févr-28 L50 OK
36 36 29896 1875 1875 26.63167 40.43205 03-janv L50 OK
37 37 26062 -3834 3834 10.91212 38.11262 03-févr L50 OK
38 38 47014 20952 20952 38.39187 32.15296 03-mars L50 LAST
39 39 56928 9914 9914 56.56955 51.64088 03-avr M50 <NA>
40 40 73152 16224 16224 65.75111 55.29208 03-mai M50 <NA>
希望这会有所帮助
Gottavianoni