使用dplyr和rollapply在数据框中滚动预测 [英] Rolling prediction in a data frame using dplyr and rollapply
本文介绍了使用dplyr和rollapply在数据框中滚动预测的处理方法，对大家解决问题具有一定的参考价值，需要的朋友们下面随着小编来一起学习吧！
问题描述
我的第一个问题:)
我的目标是:给定一个具有预测变量的数据框(每列一个预测变量/行观测值)，使用lm拟合回归，然后使用滚动窗口使用上一个观测值预测值.
My goal is: Given a data frame with predictors (each column a predictor / rows observations) fit a regression using lm and then predict the value using the last observation using a rolling window.
数据框如下:
> DfPredictor[1:40,]
Y X1 X2 X3 X4 X5
1 3.2860 192.5115 2.1275 83381 11.4360 8.7440
2 3.2650 190.1462 2.0050 88720 11.4359 8.8971
3 3.2213 192.9773 2.0500 74130 11.4623 8.8380
4 3.1991 193.7058 2.1050 73930 11.3366 8.7536
5 3.2224 193.5407 2.0275 80875 11.3534 8.7555
6 3.2000 190.6049 2.0950 86606 11.3290 8.8555
7 3.1939 191.1390 2.0975 91402 11.2960 8.8433
8 3.1971 192.2921 2.2700 88181 11.2930 8.8681
9 3.1873 194.9700 2.3300 115959 1.9477 8.5245
10 3.2182 194.5396 2.4200 134754 11.3200 8.4990
11 3.2409 194.5396 2.2025 136685 1.9649 8.4192
12 3.2112 195.1362 2.1900 136316 1.9750 8.3752
13 3.2231 193.3560 2.2475 140295 1.9691 8.3546
14 3.2015 192.9649 2.2575 139474 1.9500 8.3116
15 3.1744 194.0154 2.1900 146202 1.8476 8.2225
16 3.1646 194.4423 2.2650 142983 1.8600 8.1948
17 3.1708 194.9473 2.2425 141377 1.8522 8.2589
18 3.1675 193.9788 2.2400 141377 1.8600 8.2600
19 3.1744 194.2563 2.3000 149875 1.8718 8.2899
20 3.1410 193.4316 2.2300 129561 1.8480 8.2395
21 3.1266 191.2633 2.2550 122636 1.8440 8.2396
22 3.1486 192.0354 2.3600 130996 1.8570 8.8640
23 3.1282 194.3351 2.4825 92430 1.7849 8.1291
24 3.1214 193.5196 2.4750 94814 1.7624 8.1991
25 3.1230 193.2017 2.3725 87590 1.7660 8.2310
26 3.1182 192.1642 2.4475 87715 1.6955 8.2414
27 3.1203 191.3744 2.3775 89857 1.6539 8.2480
28 3.1156 192.2646 2.3725 92159 1.5976 8.1676
29 3.1270 192.7555 2.3675 97425 1.5896 8.1162
30 3.1154 194.0375 2.3725 87598 1.5277 8.2640
31 3.1104 192.0596 2.3850 93236 1.5132 7.9999
32 3.0846 192.2792 2.2900 94608 1.4990 8.1600
33 3.0569 193.2573 2.3050 84663 1.4715 8.2200
34 3.0893 192.7632 2.2550 67149 1.4955 7.9590
35 3.0991 192.1229 2.3050 75519 1.4280 7.9183
36 3.0879 192.1229 2.3100 76756 1.3839 7.9133
37 3.0965 192.0502 2.2175 61748 1.3130 7.8750
38 3.0655 191.2274 2.2300 41490 1.2823 7.8656
39 3.0636 191.6342 2.1925 51049 1.1492 7.7447
40 3.1097 190.9312 2.2150 21934 1.1626 7.6895
例如，使用宽度为10的滚动窗口，应该估计回归，然后预测与X1，X2，...，X5对应的"Y". 预测应包含在新列"Ypred"中.
For instance using the rolling window with width = 10 the regression should be estimate and then predict the 'Y' correspondent to the X1,X2,...,X5. The predictions should be included in a new column 'Ypred'.
有一些方法可以使用rollapply + lm/predict + mudate吗?
There's some way to do that using rollapply + lm/predict + mudate??
非常感谢！
推荐答案
使用末尾注释中的数据，并假定在宽度为10的窗口中，我们要预测最后一个Y(即第十个Y) )，然后:
Using the data in the Note at the end and assuming that in a window of width 10 we want to predict the last Y (i..e. the 10th), then:
library(zoo)
pred <- function(x) tail(fitted(lm(Y ~., as.data.frame(x))), 1)
transform(DF, pred = rollapplyr(DF, 10, pred, by.column = FALSE, fill = NA))
给予:
Y X1 X2 X3 X4 X5 pred
1 3.2860 192.5115 2.1275 83381 11.4360 8.7440 NA
2 3.2650 190.1462 2.0050 88720 11.4359 8.8971 NA
3 3.2213 192.9773 2.0500 74130 11.4623 8.8380 NA
4 3.1991 193.7058 2.1050 73930 11.3366 8.7536 NA
5 3.2224 193.5407 2.0275 80875 11.3534 8.7555 NA
6 3.2000 190.6049 2.0950 86606 11.3290 8.8555 NA
7 3.1939 191.1390 2.0975 91402 11.2960 8.8433 NA
8 3.1971 192.2921 2.2700 88181 11.2930 8.8681 NA
9 3.1873 194.9700 2.3300 115959 1.9477 8.5245 NA
10 3.2182 194.5396 2.4200 134754 11.3200 8.4990 3.219764
11 3.2409 194.5396 2.2025 136685 1.9649 8.4192 3.241614
12 3.2112 195.1362 2.1900 136316 1.9750 8.3752 3.225423
13 3.2231 193.3560 2.2475 140295 1.9691 8.3546 3.217797
14 3.2015 192.9649 2.2575 139474 1.9500 8.3116 3.205856
15 3.1744 194.0154 2.1900 146202 1.8476 8.2225 3.177928
16 3.1646 194.4423 2.2650 142983 1.8600 8.1948 3.156405
17 3.1708 194.9473 2.2425 141377 1.8522 8.2589 3.176243
18 3.1675 193.9788 2.2400 141377 1.8600 8.2600 3.177165
19 3.1744 194.2563 2.3000 149875 1.8718 8.2899 3.177211
20 3.1410 193.4316 2.2300 129561 1.8480 8.2395 3.145533
21 3.1266 191.2633 2.2550 122636 1.8440 8.2396 3.127410
22 3.1486 192.0354 2.3600 130996 1.8570 8.8640 3.148792
23 3.1282 194.3351 2.4825 92430 1.7849 8.1291 3.124913
24 3.1214 193.5196 2.4750 94814 1.7624 8.1991 3.124992
25 3.1230 193.2017 2.3725 87590 1.7660 8.2310 3.117981
26 3.1182 192.1642 2.4475 87715 1.6955 8.2414 3.117679
27 3.1203 191.3744 2.3775 89857 1.6539 8.2480 3.119898
28 3.1156 192.2646 2.3725 92159 1.5976 8.1676 3.121039
29 3.1270 192.7555 2.3675 97425 1.5896 8.1162 3.123903
30 3.1154 194.0375 2.3725 87598 1.5277 8.2640 3.119438
31 3.1104 192.0596 2.3850 93236 1.5132 7.9999 3.113963
32 3.0846 192.2792 2.2900 94608 1.4990 8.1600 3.101229
33 3.0569 193.2573 2.3050 84663 1.4715 8.2200 3.076817
34 3.0893 192.7632 2.2550 67149 1.4955 7.9590 3.083266
35 3.0991 192.1229 2.3050 75519 1.4280 7.9183 3.089377
36 3.0879 192.1229 2.3100 76756 1.3839 7.9133 3.084225
37 3.0965 192.0502 2.2175 61748 1.3130 7.8750 3.075252
38 3.0655 191.2274 2.2300 41490 1.2823 7.8656 3.063025
39 3.0636 191.6342 2.1925 51049 1.1492 7.7447 3.068808
40 3.1097 190.9312 2.2150 21934 1.1626 7.6895 3.091819
注意:可重复输入的DF是:
Lines <- " Y X1 X2 X3 X4 X5
1 3.2860 192.5115 2.1275 83381 11.4360 8.7440
2 3.2650 190.1462 2.0050 88720 11.4359 8.8971
3 3.2213 192.9773 2.0500 74130 11.4623 8.8380
4 3.1991 193.7058 2.1050 73930 11.3366 8.7536
5 3.2224 193.5407 2.0275 80875 11.3534 8.7555
6 3.2000 190.6049 2.0950 86606 11.3290 8.8555
7 3.1939 191.1390 2.0975 91402 11.2960 8.8433
8 3.1971 192.2921 2.2700 88181 11.2930 8.8681
9 3.1873 194.9700 2.3300 115959 1.9477 8.5245
10 3.2182 194.5396 2.4200 134754 11.3200 8.4990
11 3.2409 194.5396 2.2025 136685 1.9649 8.4192
12 3.2112 195.1362 2.1900 136316 1.9750 8.3752
13 3.2231 193.3560 2.2475 140295 1.9691 8.3546
14 3.2015 192.9649 2.2575 139474 1.9500 8.3116
15 3.1744 194.0154 2.1900 146202 1.8476 8.2225
16 3.1646 194.4423 2.2650 142983 1.8600 8.1948
17 3.1708 194.9473 2.2425 141377 1.8522 8.2589
18 3.1675 193.9788 2.2400 141377 1.8600 8.2600
19 3.1744 194.2563 2.3000 149875 1.8718 8.2899
20 3.1410 193.4316 2.2300 129561 1.8480 8.2395
21 3.1266 191.2633 2.2550 122636 1.8440 8.2396
22 3.1486 192.0354 2.3600 130996 1.8570 8.8640
23 3.1282 194.3351 2.4825 92430 1.7849 8.1291
24 3.1214 193.5196 2.4750 94814 1.7624 8.1991
25 3.1230 193.2017 2.3725 87590 1.7660 8.2310
26 3.1182 192.1642 2.4475 87715 1.6955 8.2414
27 3.1203 191.3744 2.3775 89857 1.6539 8.2480
28 3.1156 192.2646 2.3725 92159 1.5976 8.1676
29 3.1270 192.7555 2.3675 97425 1.5896 8.1162
30 3.1154 194.0375 2.3725 87598 1.5277 8.2640
31 3.1104 192.0596 2.3850 93236 1.5132 7.9999
32 3.0846 192.2792 2.2900 94608 1.4990 8.1600
33 3.0569 193.2573 2.3050 84663 1.4715 8.2200
34 3.0893 192.7632 2.2550 67149 1.4955 7.9590
35 3.0991 192.1229 2.3050 75519 1.4280 7.9183
36 3.0879 192.1229 2.3100 76756 1.3839 7.9133
37 3.0965 192.0502 2.2175 61748 1.3130 7.8750
38 3.0655 191.2274 2.2300 41490 1.2823 7.8656
39 3.0636 191.6342 2.1925 51049 1.1492 7.7447
40 3.1097 190.9312 2.2150 21934 1.1626 7.6895"
DF <- read.table(text = Lines, header = TRUE)
这篇关于使用dplyr和rollapply在数据框中滚动预测的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！
查看全文
