我正在阅读有关R中各小组先知奔跑的问答。 Using Prophet Package to Predict By Group in Dataframe in R
提供的答案是有用的，但不包括为先知函数添加的holidays参数。
按组运行该功能时，如何将假期data.frame传递给先知功能？
这是我的示例数据：
# time series df
ds <- as.Date(c('2016-11-01','2016-11-02','2016-11-03','2016-11-04',
'2016-11-05','2016-11-06','2016-11-07','2016-11-08',
'2016-11-09','2016-11-10','2016-11-11','2016-11-12',
'2016-11-13','2016-11-14','2016-11-15','2016-11-16',
'2016-11-17','2016-11-18','2016-11-19','2016-11-20',
'2016-11-21','2016-11-22','2016-11-23','2016-11-24',
'2016-11-25','2016-11-26','2016-11-27','2016-11-28',
'2016-11-29','2016-11-30',
'2016-11-01','2016-11-02','2016-11-03','2016-11-04',
'2016-11-05','2016-11-06','2016-11-07','2016-11-08',
'2016-11-09','2016-11-10','2016-11-11','2016-11-12',
'2016-11-13','2016-11-14','2016-11-15','2016-11-16',
'2016-11-17','2016-11-18','2016-11-19','2016-11-20',
'2016-11-21','2016-11-22','2016-11-23','2016-11-24',
'2016-11-25','2016-11-26','2016-11-27','2016-11-28'))
y <-c(15,17,18,19,20,54,67,23,12,34,12,78,34,12,3,45,67,89,12,111,123,112,14,566,345,123,567,56,87,90, 45,23,12,10,21,34,12,45,12,44,87,45,32,67,1,57,87,99,33,234,456,123,89,333,411,232,455,55)
y <- as.numeric(y)
group <- c("A","A","A","A","A","A","A","A","A","A","A","A","A","A","A",
"A","A","A","A","A","A","A","A","A","A","A","A","A","A","A",
"B","B","B","B","B","B","B","B","B","B","B","B","B","B","B",
"B","B","B","B","B","B","B","B","B","B","B","B","B")
df <- data.frame(ds,group, y)
df
ds group y
1 2016-11-01 A 15
2 2016-11-02 A 17
3 2016-11-03 A 18
4 2016-11-04 A 19
5 2016-11-05 A 20
6 2016-11-06 A 54
7 2016-11-07 A 67
8 2016-11-08 A 23
9 2016-11-09 A 12
10 2016-11-10 A 34
11 2016-11-11 A 12
12 2016-11-12 A 78
13 2016-11-13 A 34
14 2016-11-14 A 12
15 2016-11-15 A 3
16 2016-11-16 A 45
17 2016-11-17 A 67
18 2016-11-18 A 89
19 2016-11-19 A 12
20 2016-11-20 A 111
21 2016-11-21 A 123
22 2016-11-22 A 112
23 2016-11-23 A 14
24 2016-11-24 A 566
25 2016-11-25 A 345
26 2016-11-26 A 123
27 2016-11-27 A 567
28 2016-11-28 A 56
29 2016-11-29 A 87
30 2016-11-30 A 90
31 2016-11-01 B 45
32 2016-11-02 B 23
33 2016-11-03 B 12
34 2016-11-04 B 10
35 2016-11-05 B 21
36 2016-11-06 B 34
37 2016-11-07 B 12
38 2016-11-08 B 45
39 2016-11-09 B 12
40 2016-11-10 B 44
41 2016-11-11 B 87
42 2016-11-12 B 45
43 2016-11-13 B 32
44 2016-11-14 B 67
45 2016-11-15 B 1
46 2016-11-16 B 57
47 2016-11-17 B 87
48 2016-11-18 B 99
49 2016-11-19 B 33
50 2016-11-20 B 234
51 2016-11-21 B 456
52 2016-11-22 B 123
53 2016-11-23 B 89
54 2016-11-24 B 333
55 2016-11-25 B 411
56 2016-11-26 B 232
57 2016-11-27 B 455
58 2016-11-28 B 55
# holidays df
ds <- as.Date(c('2016-11-10','2016-11-23','2016-11-19','2016-11-28'))
group <- c("A","A","B","B")
holiday <- c('holiday_a', 'holiday_b','holiday_c', 'holiday_d')
holidays <- data.frame(ds,group, holiday)
holidays
ds group holiday
1 2016-11-10 A holiday_a
2 2016-11-23 A holiday_b
3 2016-11-19 B holiday_c
4 2016-11-28 B holiday_d
在将其传递给先知函数时，我尝试按假日data.frame参数进行分组，但输出不正确。所有假期a，b，b，d均由假期数据组中的组独立传递给每个组；这是不正确的。
df %>%
group_by(group) %>%
do(predict(prophet(., holidays = group_by(holidays, group)),
make_future_dataframe(prophet(.), periods = 7))) %>%
select(ds, group, yhat,holiday_a,holiday_b,holiday_c,holiday_d) %>%
as.data.frame()
ds group yhat holiday_a holiday_b holiday_c holiday_d
1 2016-11-01 A -94.6419164 0.00000 0.00000 0.00000 0.00000
2 2016-11-02 A -96.5462126 0.00000 0.00000 0.00000 0.00000
3 2016-11-03 A 90.9201486 0.00000 0.00000 0.00000 0.00000
4 2016-11-04 A 11.6291775 0.00000 0.00000 0.00000 0.00000
5 2016-11-05 A -39.6195794 0.00000 0.00000 0.00000 0.00000
6 2016-11-06 A 101.3672497 0.00000 0.00000 0.00000 0.00000
7 2016-11-07 A -27.1164855 0.00000 0.00000 0.00000 0.00000
8 2016-11-08 A -26.1950639 0.00000 0.00000 0.00000 0.00000
9 2016-11-09 A -28.0996817 0.00000 0.00000 0.00000 0.00000
10 2016-11-10 A 72.4524666 -86.91389 0.00000 0.00000 0.00000
11 2016-11-11 A 80.0750655 0.00000 0.00000 0.00000 0.00000
12 2016-11-12 A 28.8259872 0.00000 0.00000 0.00000 0.00000
13 2016-11-13 A 169.8124950 0.00000 0.00000 0.00000 0.00000
14 2016-11-14 A 41.3284385 0.00000 0.00000 0.00000 0.00000
15 2016-11-15 A 42.2498601 0.00000 0.00000 0.00000 0.00000
16 2016-11-16 A 40.3452425 0.00000 0.00000 0.00000 0.00000
17 2016-11-17 A 227.8112824 0.00000 0.00000 0.00000 0.00000
18 2016-11-18 A 148.5199899 0.00000 0.00000 0.00000 0.00000
19 2016-11-19 A 45.7585824 0.00000 0.00000 -51.51233 0.00000
20 2016-11-20 A 238.2574195 0.00000 0.00000 0.00000 0.00000
21 2016-11-21 A 109.7733629 0.00000 0.00000 0.00000 0.00000
22 2016-11-22 A 110.6947844 0.00000 0.00000 0.00000 0.00000
23 2016-11-23 A 62.5667545 0.00000 -46.22341 0.00000 0.00000
24 2016-11-24 A 296.2562062 0.00000 0.00000 0.00000 0.00000
25 2016-11-25 A 216.9649135 0.00000 0.00000 0.00000 0.00000
26 2016-11-26 A 165.7158351 0.00000 0.00000 0.00000 0.00000
27 2016-11-27 A 306.7023427 0.00000 0.00000 0.00000 0.00000
28 2016-11-28 A 126.2788594 0.00000 0.00000 0.00000 -51.93943
29 2016-11-29 A 179.1397075 0.00000 0.00000 0.00000 0.00000
30 2016-11-30 A 177.2350897 0.00000 0.00000 0.00000 0.00000
31 2016-12-01 A 364.7011292 0.00000 0.00000 0.00000 0.00000
32 2016-12-02 A 285.4098366 0.00000 0.00000 0.00000 0.00000
33 2016-12-03 A 234.1607582 0.00000 0.00000 0.00000 0.00000
34 2016-12-04 A 375.1472658 0.00000 0.00000 0.00000 0.00000
35 2016-12-05 A 246.6632091 0.00000 0.00000 0.00000 0.00000
36 2016-12-06 A 247.5846306 0.00000 0.00000 0.00000 0.00000
37 2016-12-07 A 245.6800127 0.00000 0.00000 0.00000 0.00000
38 2016-11-01 B -71.4343718 0.00000 0.00000 0.00000 0.00000
39 2016-11-02 B -77.8417828 0.00000 0.00000 0.00000 0.00000
40 2016-11-03 B 8.7292616 0.00000 0.00000 0.00000 0.00000
41 2016-11-04 B 33.1001674 0.00000 0.00000 0.00000 0.00000
42 2016-11-05 B -27.3674191 0.00000 0.00000 0.00000 0.00000
43 2016-11-06 B 72.5228028 0.00000 0.00000 0.00000 0.00000
44 2016-11-07 B 53.5127980 0.00000 0.00000 0.00000 0.00000
45 2016-11-08 B 6.8424459 0.00000 0.00000 0.00000 0.00000
46 2016-11-09 B 0.4350352 0.00000 0.00000 0.00000 0.00000
47 2016-11-10 B 43.3701494 -43.63593 0.00000 0.00000 0.00000
48 2016-11-11 B 111.3769861 0.00000 0.00000 0.00000 0.00000
49 2016-11-12 B 50.9093997 0.00000 0.00000 0.00000 0.00000
50 2016-11-13 B 150.7996217 0.00000 0.00000 0.00000 0.00000
51 2016-11-14 B 131.7896172 0.00000 0.00000 0.00000 0.00000
52 2016-11-15 B 85.1848200 0.00000 0.00000 0.00000 0.00000
53 2016-11-16 B 78.8429644 0.00000 0.00000 0.00000 0.00000
54 2016-11-17 B 165.4795640 0.00000 0.00000 0.00000 0.00000
55 2016-11-18 B 189.9160251 0.00000 0.00000 0.00000 0.00000
56 2016-11-19 B 75.9048249 0.00000 0.00000 -53.60917 0.00000
57 2016-11-20 B 229.4697708 0.00000 0.00000 0.00000 0.00000
58 2016-11-21 B 210.5253213 0.00000 0.00000 0.00000 0.00000
59 2016-11-22 B 163.9205243 0.00000 0.00000 0.00000 0.00000
60 2016-11-23 B 146.5432931 0.00000 -11.03538 0.00000 0.00000
61 2016-11-24 B 244.2152686 0.00000 0.00000 0.00000 0.00000
62 2016-11-25 B 268.6517298 0.00000 0.00000 0.00000 0.00000
63 2016-11-26 B 208.2496985 0.00000 0.00000 0.00000 0.00000
64 2016-11-27 B 308.2054755 0.00000 0.00000 0.00000 0.00000
65 2016-11-28 B 178.5735017 0.00000 0.00000 0.00000 -110.68752
66 2016-11-29 B 242.6562289 0.00000 0.00000 0.00000 0.00000
67 2016-11-30 B 236.3143734 0.00000 0.00000 0.00000 0.00000
68 2016-12-01 B 322.9509732 0.00000 0.00000 0.00000 0.00000
69 2016-12-02 B 347.3874344 0.00000 0.00000 0.00000 0.00000
70 2016-12-03 B 286.9854031 0.00000 0.00000 0.00000 0.00000
71 2016-12-04 B 386.9411801 0.00000 0.00000 0.00000 0.00000
72 2016-12-05 B 367.9967306 0.00000 0.00000 0.00000 0.00000
如何将假期data.frame传递给按组运行的先知函数，以预测每个带有相应假期的组？
答案 0 :(得分：1)
我找到了答案，使用map2 purrr函数预测了每个带有相应假期的人群
df_nested <- df %>%
group_by(group) %>%
nest()
holidays_nested <- holidays %>%
group_by(group) %>%
nest() %>%
rename(holidays = data)
df_input <- df_nested %>% left_join(holidays_nested)
df_input %>%
mutate(forecast = map2(data, holidays, ~predict(prophet(df = .x, holidays = .y), make_future_dataframe(prophet(df = .x, holidays = .y), periods = 7)))) %>%
unnest(forecast) %>%
select(ds, group, yhat,holiday_a,holiday_b,holiday_c,holiday_d) %>%
as.data.frame()
感谢那些调查问题的人。