R 根据时间阈值，将来自不同人群的多个观察结果浓缩为不同的事件

我有一个多人观看同一视频的数据集。当他们看到一只鸟进入画面时，我听到了他们的咔哒声。我们的目标是压缩他们的点击量，这样我就能知道实际发生了什么。我知道有四只鸟进入画面，所以应该有四个事件。这是我正在处理的数据的一个子集，以及点击的视觉效果，每种颜色都有不同的用户

structure(list(user_login = structure(c(7L, 7L, 7L, 7L, 4L, 4L, 
4L, 4L, 1L, 1L, 1L, 1L, 5L, 5L, 5L, 5L, 5L, 2L, 2L, 2L, 2L, 6L, 
6L, 6L, 3L), .Label = c("charles_cornellbirds", "Rachael_cornellbirds", 
"USER1125013", "USER399555", "USER413265", "USER602873", "USER947968"
), class = "factor"), time_video = c(96, 135, 209, 211, 101, 
140, 215, 216, 95, 136, 208, 209, 105, 146, 233, 234, 282, 93, 
134, 209, 209, 136, 209, 210, 101), dummy_value = c(1, 1, 1, 
1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
1)), row.names = c(8L, 15L, 21L, 22L, 34L, 42L, 51L, 52L, 69L, 
84L, 97L, 98L, 121L, 131L, 139L, 141L, 148L, 160L, 174L, 188L, 
189L, 229L, 242L, 243L, 267L), class = "data.frame")

我试着在这个类似的场景中使用寒冷，但使用后续点击之间的时间并不能让我了解4只鸟进入画面的真相。见下文：

数据

structure(list(id = c(1602L, 1549L, 1487L, 1513L, 1753L, 1712L, 
1616L, 1494L, 1564L, 1672L, 1522L, 1761L, 1722L, 1577L, 1500L, 
1578L, 1631L, 1632L, 1686L, 1687L, 1501L, 1767L, 1531L, 1768L, 
1532L, 1730L, 1732L, 1774L, 1740L), user_login = structure(c(2L, 
1L, 7L, 4L, 3L, 5L, 2L, 7L, 1L, 6L, 4L, 3L, 5L, 1L, 7L, 1L, 2L, 
2L, 6L, 6L, 7L, 3L, 4L, 3L, 4L, 5L, 5L, 3L, 5L), .Label = c("USER_EXPERT", 
"USER_EXPERT2", "USER1125013", "USER399555", "USER413265", "USER602873", 
"USER947968"), class = "factor"), time_in_utc = structure(c(118L, 
68L, 8L, 33L, 255L, 220L, 131L, 15L, 80L, 182L, 41L, 262L, 228L, 
93L, 20L, 94L, 145L, 145L, 195L, 196L, 21L, 268L, 50L, 269L, 
51L, 233L, 234L, 275L, 242L), .Label = c("10/29/2019 17:35:21", 
"10/29/2019 18:01:49", "10/29/2019 18:01:53", "10/29/2019 18:02:01", 
"10/29/2019 18:02:14", "10/29/2019 18:03:09", "10/29/2019 18:03:18", 
"10/29/2019 18:03:25", "10/29/2019 18:03:30", "10/29/2019 18:03:32", 
"10/29/2019 18:03:38", "10/29/2019 18:03:41", "10/29/2019 18:04:00", 
"10/29/2019 18:04:03", "10/29/2019 18:04:04", "10/29/2019 18:04:32", 
"10/29/2019 18:04:33", "10/29/2019 18:04:50", "10/29/2019 18:05:17", 
"10/29/2019 18:05:18", "10/29/2019 18:05:20", "10/29/2019 18:05:38", 
"10/29/2019 18:05:39", "10/29/2019 18:05:45", "10/29/2019 18:54:24", 
"10/29/2019 18:54:30", "10/29/2019 18:54:39", "10/29/2019 18:54:45", 
"10/29/2019 18:55:01", "10/29/2019 18:55:29", "10/29/2019 18:55:55", 
"10/29/2019 18:56:04", "10/29/2019 18:56:11", "10/29/2019 18:56:12", 
"10/29/2019 18:56:18", "10/29/2019 18:56:23", "10/29/2019 18:56:26", 
"10/29/2019 18:56:39", "10/29/2019 18:56:45", "10/29/2019 18:56:47", 
"10/29/2019 18:56:50", "10/29/2019 18:56:55", "10/29/2019 18:56:59", 
"10/29/2019 18:57:10", "10/29/2019 18:57:12", "10/29/2019 18:57:14", 
"10/29/2019 18:57:23", "10/29/2019 18:57:36", "10/29/2019 18:57:37", 
"10/29/2019 18:58:05", "10/29/2019 18:58:06", "10/29/2019 18:58:19", 
"10/29/2019 18:58:25", "10/29/2019 18:58:37", "10/29/2019 18:58:39", 
"10/29/2019 18:58:52", "10/29/2019 18:58:54", "10/29/2019 18:59:17", 
"10/29/2019 18:59:41", "10/30/2019 13:43:32", "10/30/2019 13:45:16", 
"10/30/2019 13:45:20", "10/30/2019 13:45:24", "10/30/2019 13:45:41", 
"10/30/2019 13:46:10", "10/30/2019 13:46:35", "10/30/2019 13:46:44", 
"10/30/2019 13:46:51", "10/30/2019 13:46:52", "10/30/2019 13:46:53", 
"10/30/2019 13:46:58", "10/30/2019 13:47:03", "10/30/2019 13:47:06", 
"10/30/2019 13:47:08", "10/30/2019 13:47:19", "10/30/2019 13:47:22", 
"10/30/2019 13:47:25", "10/30/2019 13:47:26", "10/30/2019 13:47:31", 
"10/30/2019 13:47:32", "10/30/2019 13:47:34", "10/30/2019 13:47:35", 
"10/30/2019 13:47:39", "10/30/2019 13:47:41", "10/30/2019 13:47:48", 
"10/30/2019 13:47:49", "10/30/2019 13:47:52", "10/30/2019 13:47:53", 
"10/30/2019 13:48:04", "10/30/2019 13:48:15", "10/30/2019 13:48:16", 
"10/30/2019 13:48:18", "10/30/2019 13:48:44", "10/30/2019 13:48:45", 
"10/30/2019 13:48:46", "10/30/2019 13:48:49", "10/30/2019 13:48:59", 
"10/30/2019 13:49:06", "10/30/2019 13:49:11", "10/30/2019 13:49:17", 
"10/30/2019 13:49:20", "10/30/2019 13:49:30", "10/30/2019 13:49:33", 
"10/30/2019 13:49:35", "10/30/2019 13:49:37", "10/30/2019 13:49:44", 
"10/30/2019 13:50:36", "10/31/2019 13:25:44", "10/31/2019 15:45:07", 
"10/31/2019 15:45:32", "10/31/2019 15:46:52", "10/31/2019 15:46:55", 
"10/31/2019 15:47:00", "10/31/2019 15:47:17", "10/31/2019 15:47:44", 
"10/31/2019 15:48:12", "10/31/2019 15:48:18", "10/31/2019 15:48:25", 
"10/31/2019 15:48:27", "10/31/2019 15:48:28", "10/31/2019 15:48:32", 
"10/31/2019 15:48:34", "10/31/2019 15:48:37", "10/31/2019 15:48:39", 
"10/31/2019 15:48:40", "10/31/2019 15:48:41", "10/31/2019 15:48:55", 
"10/31/2019 15:48:56", "10/31/2019 15:48:59", "10/31/2019 15:49:01", 
"10/31/2019 15:49:06", "10/31/2019 15:49:07", "10/31/2019 15:49:10", 
"10/31/2019 15:49:15", "10/31/2019 15:49:17", "10/31/2019 15:49:23", 
"10/31/2019 15:49:24", "10/31/2019 15:49:25", "10/31/2019 15:49:28", 
"10/31/2019 15:49:39", "10/31/2019 15:49:51", "10/31/2019 15:49:53", 
"10/31/2019 15:49:59", "10/31/2019 15:50:20", "10/31/2019 15:50:21", 
"10/31/2019 15:50:23", "10/31/2019 15:50:24", "10/31/2019 15:50:34", 
"10/31/2019 15:50:41", "10/31/2019 15:50:47", "10/31/2019 15:50:53", 
"10/31/2019 15:50:55", "10/31/2019 15:51:07", "10/31/2019 15:51:09", 
"10/31/2019 15:51:10", "10/31/2019 15:51:18", "10/31/2019 15:51:19", 
"10/31/2019 15:51:31", "10/31/2019 15:52:00", "10/31/2019 17:02:57", 
"10/31/2019 17:03:10", "10/31/2019 17:03:15", "10/31/2019 17:03:19", 
"10/31/2019 17:03:35", "10/31/2019 17:04:03", "10/31/2019 17:04:29", 
"10/31/2019 17:04:38", "10/31/2019 17:04:46", "10/31/2019 17:04:49", 
"10/31/2019 17:04:51", "10/31/2019 17:04:53", "10/31/2019 17:04:55", 
"10/31/2019 17:04:58", "10/31/2019 17:05:00", "10/31/2019 17:05:01", 
"10/31/2019 17:05:02", "10/31/2019 17:05:14", "10/31/2019 17:05:17", 
"10/31/2019 17:05:20", "10/31/2019 17:05:21", "10/31/2019 17:05:25", 
"10/31/2019 17:05:26", "10/31/2019 17:05:29", "10/31/2019 17:05:33", 
"10/31/2019 17:05:34", "10/31/2019 17:05:35", "10/31/2019 17:05:41", 
"10/31/2019 17:05:44", "10/31/2019 17:05:47", "10/31/2019 17:05:58", 
"10/31/2019 17:06:00", "10/31/2019 17:06:09", "10/31/2019 17:06:11", 
"10/31/2019 17:06:13", "10/31/2019 17:06:39", "10/31/2019 17:06:40", 
"10/31/2019 17:06:41", "10/31/2019 17:06:42", "10/31/2019 17:06:52", 
"10/31/2019 17:07:00", "10/31/2019 17:07:06", "10/31/2019 17:07:07", 
"10/31/2019 17:07:12", "10/31/2019 17:07:14", "10/31/2019 17:07:24", 
"10/31/2019 17:07:25", "10/31/2019 17:07:30", "10/31/2019 17:07:37", 
"10/31/2019 17:07:39", "10/31/2019 17:07:51", "10/31/2019 17:08:19", 
"10/31/2019 20:52:14", "11/01/2019 17:37:53", "11/01/2019 17:37:56", 
"11/01/2019 17:38:14", "11/01/2019 17:38:17", "11/01/2019 17:38:34", 
"11/01/2019 17:39:26", "11/01/2019 17:39:34", "11/01/2019 17:39:41", 
"11/01/2019 17:39:46", "11/01/2019 17:39:47", "11/01/2019 17:39:49", 
"11/01/2019 17:39:54", "11/01/2019 17:40:04", "11/01/2019 17:40:12", 
"11/01/2019 17:40:16", "11/01/2019 17:40:22", "11/01/2019 17:40:29", 
"11/01/2019 17:40:53", "11/01/2019 17:41:01", "11/01/2019 17:41:25", 
"11/01/2019 17:41:49", "11/01/2019 17:41:50", "11/01/2019 17:42:03", 
"11/01/2019 17:42:10", "11/01/2019 17:42:16", "11/01/2019 17:42:17", 
"11/01/2019 17:42:19", "11/01/2019 17:42:26", "11/01/2019 17:42:32", 
"11/01/2019 17:42:38", "11/01/2019 17:42:45", "11/01/2019 17:43:27", 
"11/01/2019 23:01:41", "11/01/2019 23:01:51", "11/01/2019 23:02:02", 
"11/01/2019 23:02:05", "11/01/2019 23:02:16", "11/01/2019 23:02:23", 
"11/01/2019 23:02:25", "11/01/2019 23:02:48", "11/01/2019 23:03:19", 
"11/01/2019 23:03:25", "11/01/2019 23:03:32", "11/01/2019 23:03:34", 
"11/01/2019 23:03:35", "11/01/2019 23:03:40", "11/01/2019 23:03:45", 
"11/01/2019 23:03:49", "11/01/2019 23:03:50", "11/01/2019 23:04:12", 
"11/01/2019 23:04:17", "11/01/2019 23:04:20", "11/01/2019 23:04:36", 
"11/01/2019 23:04:45", "11/01/2019 23:04:59", "11/01/2019 23:05:24", 
"11/01/2019 23:05:26", "11/01/2019 23:05:27", "11/01/2019 23:05:40", 
"11/01/2019 23:05:47", "11/01/2019 23:05:57", "11/01/2019 23:05:59", 
"11/01/2019 23:06:15", "11/01/2019 23:06:16", "11/01/2019 23:06:26", 
"11/01/2019 23:06:38", "11/01/2019 23:07:04"), class = "factor"), 
    time = structure(c(59L, 10L, 192L, 217L, 255L, 161L, 72L, 
    199L, 22L, 123L, 225L, 262L, 169L, 35L, 204L, 36L, 86L, 86L, 
    136L, 137L, 205L, 268L, 234L, 269L, 235L, 174L, 175L, 275L, 
    183L), .Label = c("13:25:44", "13:43:32", "13:45:16", "13:45:20", 
    "13:45:24", "13:45:41", "13:46:10", "13:46:35", "13:46:44", 
    "13:46:51", "13:46:52", "13:46:53", "13:46:58", "13:47:03", 
    "13:47:06", "13:47:08", "13:47:19", "13:47:22", "13:47:25", 
    "13:47:26", "13:47:31", "13:47:32", "13:47:34", "13:47:35", 
    "13:47:39", "13:47:41", "13:47:48", "13:47:49", "13:47:52", 
    "13:47:53", "13:48:04", "13:48:15", "13:48:16", "13:48:18", 
    "13:48:44", "13:48:45", "13:48:46", "13:48:49", "13:48:59", 
    "13:49:06", "13:49:11", "13:49:17", "13:49:20", "13:49:30", 
    "13:49:33", "13:49:35", "13:49:37", "13:49:44", "13:50:36", 
    "15:45:07", "15:45:32", "15:46:52", "15:46:55", "15:47:00", 
    "15:47:17", "15:47:44", "15:48:12", "15:48:18", "15:48:25", 
    "15:48:27", "15:48:28", "15:48:32", "15:48:34", "15:48:37", 
    "15:48:39", "15:48:40", "15:48:41", "15:48:55", "15:48:56", 
    "15:48:59", "15:49:01", "15:49:06", "15:49:07", "15:49:10", 
    "15:49:15", "15:49:17", "15:49:23", "15:49:24", "15:49:25", 
    "15:49:28", "15:49:39", "15:49:51", "15:49:53", "15:49:59", 
    "15:50:20", "15:50:21", "15:50:23", "15:50:24", "15:50:34", 
    "15:50:41", "15:50:47", "15:50:53", "15:50:55", "15:51:07", 
    "15:51:09", "15:51:10", "15:51:18", "15:51:19", "15:51:31", 
    "15:52:00", "17:02:57", "17:03:10", "17:03:15", "17:03:19", 
    "17:03:35", "17:04:03", "17:04:29", "17:04:38", "17:04:46", 
    "17:04:49", "17:04:51", "17:04:53", "17:04:55", "17:04:58", 
    "17:05:00", "17:05:01", "17:05:02", "17:05:14", "17:05:17", 
    "17:05:20", "17:05:21", "17:05:25", "17:05:26", "17:05:29", 
    "17:05:33", "17:05:34", "17:05:35", "17:05:41", "17:05:44", 
    "17:05:47", "17:05:58", "17:06:00", "17:06:09", "17:06:11", 
    "17:06:13", "17:06:39", "17:06:40", "17:06:41", "17:06:42", 
    "17:06:52", "17:07:00", "17:07:06", "17:07:07", "17:07:12", 
    "17:07:14", "17:07:24", "17:07:25", "17:07:30", "17:07:37", 
    "17:07:39", "17:07:51", "17:08:19", "17:35:21", "17:37:53", 
    "17:37:56", "17:38:14", "17:38:17", "17:38:34", "17:39:26", 
    "17:39:34", "17:39:41", "17:39:46", "17:39:47", "17:39:49", 
    "17:39:54", "17:40:04", "17:40:12", "17:40:16", "17:40:22", 
    "17:40:29", "17:40:53", "17:41:01", "17:41:25", "17:41:49", 
    "17:41:50", "17:42:03", "17:42:10", "17:42:16", "17:42:17", 
    "17:42:19", "17:42:26", "17:42:32", "17:42:38", "17:42:45", 
    "17:43:27", "18:01:49", "18:01:53", "18:02:01", "18:02:14", 
    "18:03:09", "18:03:18", "18:03:25", "18:03:30", "18:03:32", 
    "18:03:38", "18:03:41", "18:04:00", "18:04:03", "18:04:04", 
    "18:04:32", "18:04:33", "18:04:50", "18:05:17", "18:05:18", 
    "18:05:20", "18:05:38", "18:05:39", "18:05:45", "18:54:24", 
    "18:54:30", "18:54:39", "18:54:45", "18:55:01", "18:55:29", 
    "18:55:55", "18:56:04", "18:56:11", "18:56:12", "18:56:18", 
    "18:56:23", "18:56:26", "18:56:39", "18:56:45", "18:56:47", 
    "18:56:50", "18:56:55", "18:56:59", "18:57:10", "18:57:12", 
    "18:57:14", "18:57:23", "18:57:36", "18:57:37", "18:58:05", 
    "18:58:06", "18:58:19", "18:58:25", "18:58:37", "18:58:39", 
    "18:58:52", "18:58:54", "18:59:17", "18:59:41", "20:52:14", 
    "23:01:41", "23:01:51", "23:02:02", "23:02:05", "23:02:16", 
    "23:02:23", "23:02:25", "23:02:48", "23:03:19", "23:03:25", 
    "23:03:32", "23:03:34", "23:03:35", "23:03:40", "23:03:45", 
    "23:03:49", "23:03:50", "23:04:12", "23:04:17", "23:04:20", 
    "23:04:36", "23:04:45", "23:04:59", "23:05:24", "23:05:26", 
    "23:05:27", "23:05:40", "23:05:47", "23:05:57", "23:05:59", 
    "23:06:15", "23:06:16", "23:06:26", "23:06:38", "23:07:04"
    ), class = "factor"), study_name = structure(c(1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 
    1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L, 1L), .Label = "Collecting Data At The Panama Fruit Feeder", class = "factor"), 
    observation_name = structure(c(3L, 3L, 3L, 3L, 3L, 3L, 3L, 
    3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 3L, 
    3L, 3L, 3L, 3L, 3L, 3L, 3L), .Label = c("activity", "session_status", 
    "species"), class = "factor"), observation_value = structure(c(6L, 
    6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 
    6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L, 6L), .Label = c("Aggressive behavior", 
    "Chestnut-headed Oropendola", "Clay-colored Thrush", "Gray-cowled Wood-Rail", 
    "Gray-headed Chachalaca", "Rufous Motmot", "start", "stop"
    ), class = "factor"), delete_flag = c(0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 0, 
    0, 0, 0), time_video = c(93, 95, 96, 101, 101, 105, 134, 
    135, 136, 136, 140, 141, 146, 208, 209, 209, 209, 209, 209, 
    210, 211, 213, 215, 215, 216, 233, 234, 264, 282), date_time = structure(c(1572551305, 
    1572457611, 1572386605, 1572389771, 1572663812, 1572644381, 
    1572551346, 1572386644, 1572457652, 1572555926, 1572389810, 
    1572663852, 1572644422, 1572457724, 1572386718, 1572457725, 
    1572551421, 1572551421, 1572555999, 1572556000, 1572386720, 
    1572663924, 1572389885, 1572663926, 1572389886, 1572644509, 
    1572644510, 1572663975, 1572644558), class = c("POSIXct", 
    "POSIXt"), tzone = ""), date = structure(c(18200, 18199, 
    18198, 18198, 18202, 18201, 18200, 18198, 18199, 18200, 18198, 
    18202, 18201, 18199, 18198, 18199, 18200, 18200, 18200, 18200, 
    18198, 18202, 18198, 18202, 18198, 18201, 18201, 18202, 18201
    ), class = "Date"), dummy_value = c(1, 1, 1, 1, 1, 1, 1, 
    1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 1, 
    1, 1, 1)), row.names = c(160L, 69L, 8L, 34L, 267L, 121L, 
174L, 15L, 84L, 229L, 42L, 275L, 131L, 97L, 21L, 98L, 188L, 189L, 
242L, 243L, 22L, 281L, 51L, 282L, 52L, 139L, 141L, 288L, 148L
), class = "data.frame")

我试过的代码没有得到我想要的结果

v <- subset2 %>%
  dplyr::arrange(time_video) # can just do by time_video and not date because right now the video is canned 


v2 <- v %>%
  mutate(time_since_last = (time_video - lag(time_video, default = first(time_video)))) %>%
  mutate(group = 1 + cumsum(time_since_last > 20)) %>% # How many times was there a 20 second gap? Each new one increments "group"
    # this assigns each contiguous set of events to a single group
    # each new group = has there been break of __ or more? 
    # cumsum() counts how many cumulative times that has happened in the rows leading up to that point
    # ex: everything with group = 1 is before that has happened yet, everything in group = 2 is after that has happened once, 
    # everything in group = 3 is after that has happened twice, etc. 
  group_by(group) %>%
  summarize(first = min(time_video), # or first(date.time) if sorted
            last  = max(time_video), # or last(date.time) if sorted
            count = n()) %>%
  mutate(time = last-first)

ggplot() + 
  geom_rect(data = v2, aes(xmin=first, xmax=last, ymin=0, ymax=count), alpha=0.8, color = "steelblue", fill="steelblue")  + # >60 secs
  geom_text(data = v2, aes(x=first, y=count, label = count), vjust = -0.4, hjust = 0.3, color = "steelblue", size = 5) + 
  theme_minimal() + 
  labs(y = "Number of Observations", x = "Time") + 
  theme(text = element_text(size=20)) + 
  geom_point(data=subset(subset2, user_login == "USER_EXPERT"), mapping = aes(x=time_video, y=dummy_value), size=5, shape=21, fill="red")

它有4个事件，是的，但不是正确的。红色的狗是专家点击，而蓝色的矩形不能很好地反映这一点。我试着改变阈值20，看看我是否能更接近专家点击，但我认为我分组的方式不是很好。

---

我不确定这是否能解决问题，但OP能试试吗？我主要的怀疑是没有足够的数据来包含多个组

library(dplyr)
library(lubridate)

subset2 %>%
  arrange(user_login, date_time) %>%
  group_by(user_login) %>%
  mutate(time_since_last = (date_time - lag(date_time, default = first(date_time))) / dminutes(1)) %>%
  mutate(group = 1 + cumsum(time_since_last > 20)) %>% 
  group_by(user_login, group) %>%
  summarize(count = n()) %>%
  ggplot(aes(x = group, y = count, fill = user_login)) +
  geom_col(position = position_dodge()) +
  theme_minimal() + 
  labs(y = "Number of Observations", x = "Time")

---

这真是太好了！它允许我总结每个用户的观察次数。这很难解释，所以我想我错过了我要做的事情。我想浓缩所有用户的观察结果，这样我就只剩下点击了。我将添加一个编辑，以帮助它更清晰。