Question

我一直在尝试各种时间序列插补模型以使用imputeTS来插补丢失的时间序列，但是测试结果相当差。由于它看起来像季节性的时间序列数据，因此我考虑过简单地用几个月内的中位数或平均值来估算缺失值。

数据如下：

> head(data, 10)
     v1 v2  v3  v4
1  1908  1 118  10
2  1908  2 138  33
3  1908  3 128  17
4  1908  4  NA  60
5  1908  5  NA 114
6  1908  6  72 124
7  1908  7  NA  44
8  1908  8  70 134
9  1908  9  58 121
10 1908 10  42 117

我计算了中位数：

data %>%
  as_tibble() %>%
  group_by(v2) %>%
  summarise(imp_v3 = median(v3, na.rm = TRUE),
            imp_v4 = median(v4, na.rm = TRUE))

哪个给：

# A tibble: 12 x 3
      v2 imp_v3 imp_v4
   <int>  <dbl>  <dbl>
 1     1  126     23  
 2     2  132     27.5
 3     3  138     33  
 4     4   22.5   76  
 5     5   42.5  102. 
 6     6   72    127  
 7     7   82     50  
 8     8   80     51  
 9     9   57    121  
10    10   24     98  
11    11  145     71  
12    12  130.    31

我现在要做的就是将v1和v2的缺失值与相应的medians一起估算。

v3中所有丢失的值（在1中具有v2的值都将替换为126。 v4中v2列中有5个值的所有缺失值都将替换为102.。

v1中的数字是年份，v2中的数字是月份。

数据：

data <- structure(list(v1 = c(1908L, 1908L, 1908L, 1908L, 1908L, 1908L, 
1908L, 1908L, 1908L, 1908L, 1908L, 1908L, 1909L, 1909L, 1909L, 
1909L, 1909L, 1909L, 1909L, 1909L, 1909L, 1909L, 1909L, 1909L, 
1910L, 1910L, 1910L, 1910L, 1910L, 1910L, 1910L, 1910L, 1910L, 
1910L, 1910L, 1910L, 1911L, 1911L, 1911L, 1911L, 1911L, 1911L, 
1911L, 1911L, 1911L, 1911L, 1911L, 1911L, 1912L, 1912L, 1912L, 
1912L, 1912L, 1912L, 1912L, 1912L, 1912L, 1912L, 1912L, 1912L, 
1913L, 1913L, 1913L, 1913L, 1913L, 1913L, 1913L, 1913L, 1913L, 
1913L, 1913L, 1913L, 1914L, 1914L, 1914L, 1914L, 1914L, 1914L, 
1914L, 1914L, 1914L, 1914L, 1914L, 1914L, 1915L, 1915L, 1915L, 
1915L, 1915L, 1915L, 1915L, 1915L, 1915L, 1915L, 1915L, 1915L, 
1916L, 1916L, 1916L, 1916L, 1916L, 1916L, 1916L, 1916L, 1916L, 
1916L, 1916L, 1916L, 1917L, 1917L, 1917L, 1917L, 1917L, 1917L, 
1917L, 1917L, 1917L, 1917L, 1917L, 1917L, 1918L, 1918L, 1918L, 
1918L, 1918L, 1918L, 1918L, 1918L, 1918L, 1918L, 1918L, 1918L, 
1919L, 1919L, 1919L, 1919L, 1919L, 1919L, 1919L, 1919L, 1919L, 
1919L, 1919L, 1919L, 1920L, 1920L, 1920L, 1920L, 1920L, 1920L, 
1920L, 1920L, 1920L, 1920L, 1920L, 1920L, 1921L, 1921L, 1921L, 
1921L, 1921L, 1921L, 1921L, 1921L, 1921L, 1921L, 1921L, 1921L, 
1922L, 1922L, 1922L, 1922L, 1922L, 1922L, 1922L, 1922L, 1922L, 
1922L, 1922L, 1922L, 1923L, 1923L, 1923L, 1923L, 1923L, 1923L, 
1923L, 1923L, 1923L, 1923L, 1923L, 1923L, 1924L, 1924L, 1924L, 
1924L, 1924L, 1924L, 1924L, 1924L, 1924L, 1924L, 1924L, 1924L, 
1925L, 1925L, 1925L, 1925L, 1925L, 1925L, 1925L, 1925L, 1925L, 
1925L, 1925L, 1925L, 1926L, 1926L, 1926L, 1926L, 1926L, 1926L, 
1926L, 1926L, 1926L, 1926L, 1926L, 1926L, 1927L, 1927L, 1927L, 
1927L, 1927L, 1927L, 1927L, 1927L, 1927L, 1927L, 1927L, 1927L, 
1928L, 1928L, 1928L, 1928L, 1928L, 1928L, 1928L, 1928L, 1928L, 
1928L, 1928L, 1928L, 1929L, 1929L, 1929L, 1930L, 1930L, 1930L, 
1930L, 1930L, 1930L, 1930L, 1930L, 1930L, 1930L, 1930L, 1930L, 
1931L, 1931L, 1931L, 1931L, 1931L, 1931L, 1931L, 1931L, 1931L, 
1931L, 1931L, 1931L, 1932L, 1932L, 1932L, 1932L, 1932L, 1932L, 
1932L, 1932L, 1932L, 1932L, 1932L, 1932L, 1933L, 1933L, 1933L, 
1933L, 1933L, 1933L, 1933L, 1933L, 1933L, 1933L, 1933L, 1933L, 
1934L, 1934L, 1934L, 1934L, 1934L, 1934L, 1934L, 1934L, 1934L, 
1934L, 1934L, 1934L, 1935L, 1935L, 1935L, 1935L, 1935L, 1935L, 
1935L, 1935L, 1935L, 1935L, 1935L, 1935L, 1936L, 1936L, 1936L, 
1936L, 1936L, 1936L, 1936L, 1936L, 1936L, 1936L, 1936L, 1936L, 
1937L, 1937L, 1937L, 1937L, 1937L, 1937L, 1937L, 1937L, 1937L, 
1937L, 1937L, 1937L, 1938L, 1938L, 1938L, 1938L, 1938L, 1938L, 
1938L, 1938L, 1938L, 1938L, 1938L, 1938L, 1939L, 1939L, 1939L, 
1939L, 1939L, 1939L, 1939L, 1939L, 1939L, 1939L, 1939L, 1939L, 
1940L, 1940L, 1940L, 1940L, 1940L, 1940L, 1940L, 1940L, 1940L, 
1940L, 1940L, 1940L, 1941L, 1941L, 1941L, 1941L, 1941L, 1941L, 
1941L, 1941L, 1941L, 1941L, 1941L, 1941L, 1942L, 1942L, 1942L, 
1942L, 1942L, 1942L, 1942L, 1942L, 1942L, 1942L, 1942L, 1942L, 
1943L, 1943L, 1943L, 1943L, 1943L, 1943L, 1943L, 1943L, 1943L, 
1943L, 1943L, 1943L, 1944L, 1944L, 1944L, 1944L, 1944L, 1944L, 
1944L, 1944L, 1944L, 1944L, 1944L, 1944L, 1945L, 1945L, 1945L, 
1945L, 1945L, 1945L, 1945L, 1945L), v2 = c(1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 
5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 
8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 
11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 
1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 
4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 
7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 
10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 
6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 
9L, 10L, 11L, 12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 
12L, 1L, 2L, 3L, 4L, 5L, 6L, 7L, 8L, 9L, 10L, 11L, 12L, 1L, 2L, 
3L, 4L, 5L, 6L, 7L, 8L), v3 = c(118L, 138L, 128L, NA, NA, 72L, 
NA, 70L, 58L, 42L, 159L, 125L, 118L, 122L, 123L, 22L, 43L, 45L, 
68L, 82L, 41L, 29L, 140L, 120L, 119L, NA, 156L, 161L, 39L, 72L, 
65L, 75L, 45L, 26L, 122L, 145L, 125L, 135L, 130L, 3L, 58L, 73L, 
103L, 101L, 68L, 16L, 144L, 147L, NA, 132L, NA, 24L, NA, NA, 
77L, 46L, 34L, 18L, 148L, 154L, 121L, NA, 148L, 9L, 43L, 69L, 
70L, 79L, 64L, 27L, 4L, 134L, 126L, 158L, NA, 37L, 34L, 78L, 
86L, 91L, 65L, NA, 156L, 130L, 121L, 129L, 144L, 12L, 36L, 80L, 
NA, 77L, 63L, NA, 125L, NA, 160L, 121L, 112L, 16L, 44L, 40L, 
NA, 96L, NA, NA, 153L, 112L, 105L, 107L, 122L, 148L, 58L, NA, 
92L, 76L, 68L, 5L, 159L, 113L, 127L, 145L, 147L, NA, 64L, 61L, 
NA, NA, NA, 20L, 144L, 153L, NA, 106L, 121L, 4L, 68L, 72L, 67L, 
83L, NA, 14L, 117L, 137L, 139L, 148L, 4L, 160L, 44L, 70L, 58L, 
59L, 54L, 27L, 152L, 128L, 154L, 130L, 162L, 17L, 47L, 80L, 104L, 
NA, 72L, 57L, 127L, 155L, 116L, 127L, 129L, 151L, 59L, 65L, NA, 
56L, 44L, NA, 153L, 140L, 147L, 128L, 155L, 161L, 17L, NA, 95L, 
70L, 47L, 18L, 122L, NA, 126L, 119L, 132L, 160L, 35L, 62L, 75L, 
62L, NA, 24L, 154L, 155L, 138L, 134L, 138L, 6L, 42L, 83L, 98L, 
82L, 39L, 30L, 125L, 114L, 133L, 147L, 150L, 23L, NA, 67L, 97L, 
87L, 66L, 8L, 142L, 130L, 133L, 132L, 158L, 5L, 36L, NA, 81L, 
74L, 45L, 31L, 145L, NA, NA, 142L, 143L, 14L, 36L, 49L, 82L, 
80L, 54L, 24L, 161L, 124L, 21L, NA, 137L, 141L, 109L, NA, 162L, 
34L, 81L, 70L, 80L, 52L, 26L, 150L, 126L, 122L, 119L, 130L, NA, 
41L, 62L, 75L, NA, 39L, 22L, 158L, 138L, 153L, 125L, 141L, 158L, 
28L, 73L, 84L, 90L, 54L, 16L, 147L, 138L, 112L, NA, 4L, 23L, 
51L, 89L, 100L, 99L, 83L, 23L, 141L, 111L, 138L, 136L, 138L, 
10L, 48L, 77L, 104L, 80L, 79L, 26L, NA, 157L, 131L, 140L, 154L, 
10L, 33L, 81L, 99L, NA, 60L, 20L, 152L, 113L, 118L, 110L, 153L, 
NA, 41L, 76L, 78L, 88L, 66L, 25L, 144L, 142L, 135L, 134L, 118L, 
16L, 50L, 66L, 90L, NA, 59L, 28L, 142L, 113L, 143L, 133L, 25L, 
15L, 38L, 69L, 74L, 87L, 61L, 24L, 11L, 127L, 121L, 141L, 144L, 
NA, 45L, 76L, 78L, 87L, 62L, 12L, 3L, 119L, 1L, NA, 154L, 7L, 
64L, 102L, 78L, 79L, 53L, 19L, 151L, 126L, 2L, 115L, NA, 146L, 
23L, 85L, 103L, 74L, 71L, NA, 151L, 140L, 94L, 93L, 136L, 22L, 
40L, 79L, 82L, 85L, 56L, 28L, 141L, 152L, NA, 150L, 162L, 35L, 
56L, 72L, 92L, 75L, 51L, 31L, 149L, 126L, 147L, NA, 146L, 26L, 
40L, 55L, 82L, 88L, 47L, 13L, 149L, 131L, 108L, 159L, 17L, 32L, 
49L, 72L, 95L, 83L), v4 = c(10L, 33L, 17L, 60L, 114L, 124L, 44L, 
134L, 121L, 117L, 73L, NA, 15L, 3L, 3L, 72L, 86L, 112L, 42L, 
41L, 118L, 106L, 31L, 18L, 5L, 29L, NA, 63L, NA, 133L, 135L, 
52L, 122L, 110L, 4L, 75L, NA, 27L, 33L, 78L, 116L, 127L, 51L, 
NA, 123L, 94L, 71L, 72L, 15L, NA, 74L, 73L, 103L, 136L, 51L, 
129L, 114L, 81L, NA, 63L, 25L, 26L, 60L, 77L, 105L, 132L, 38L, 
NA, 37L, NA, 89L, 62L, 21L, 74L, 65L, 82L, 96L, 128L, 49L, 45L, 
121L, 109L, NA, 31L, 27L, 22L, 29L, 74L, 87L, 119L, 37L, 36L, 
126L, 89L, 6L, 26L, 82L, 19L, NA, 70L, 100L, 114L, 45L, 54L, 
126L, 98L, 80L, 6L, 5L, 12L, NA, 22L, 103L, 131L, 44L, 53L, 37L, 
75L, NA, 1L, 16L, 68L, 59L, 68L, 110L, NA, 39L, 48L, 118L, 98L, 
31L, 76L, 15L, 10L, NA, NA, 110L, 131L, NA, 39L, 120L, 70L, 16L, 
59L, NA, NA, 70L, 79L, 107L, 127L, 38L, 134L, 122L, 105L, 83L, 
31L, 77L, 33L, NA, NA, 96L, 123L, 53L, 49L, 130L, 125L, 23L, 
72L, NA, 19L, 27L, 21L, NA, 123L, 133L, 136L, 121L, 87L, 72L, 
72L, 69L, 33L, 67L, 71L, 85L, 120L, NA, 41L, NA, 102L, 19L, 19L, 
25L, NA, 9L, NA, 107L, 131L, 43L, 34L, 36L, 104L, 83L, 86L, 61L, 
28L, 32L, 73L, 109L, 129L, 54L, 51L, 112L, 101L, 2L, 8L, 29L, 
74L, 70L, 87L, 84L, 120L, NA, 44L, NA, 80L, 61L, 30L, 33L, 20L, 
74L, 77L, 93L, 111L, 50L, 51L, 129L, 102L, 67L, 5L, 23L, 62L, 
62L, 72L, 94L, 115L, 47L, 44L, 122L, 98L, 86L, 1L, 90L, 64L, 
63L, 27L, 5L, 24L, 81L, 95L, 133L, NA, NA, 136L, 107L, NA, 29L, 
14L, 24L, 7L, 77L, 105L, 135L, 52L, 136L, 120L, 84L, 88L, 63L, 
70L, 18L, 26L, 63L, NA, 125L, 56L, 56L, 124L, 85L, NA, 75L, 3L, 
24L, 60L, 86L, 114L, 136L, 56L, 57L, NA, NA, 71L, 1L, 29L, 29L, 
25L, 78L, 102L, 132L, 58L, 41L, 34L, 111L, 76L, 90L, 33L, 66L, 
72L, 78L, 86L, 37L, 52L, 48L, 129L, 98L, NA, NA, 19L, 9L, 71L, 
60L, 101L, 133L, 50L, NA, 46L, 96L, NA, 59L, 31L, 30L, 6L, 88L, 
113L, NA, 56L, 51L, 126L, 107L, 61L, 17L, NA, 62L, 91L, 74L, 
100L, 35L, 43L, 52L, 134L, 108L, 97L, 26L, 23L, 63L, 66L, 80L, 
NA, 125L, 49L, 56L, 40L, 86L, 88L, 28L, 13L, 11L, 33L, 75L, 109L, 
40L, 43L, 39L, 119L, 99L, 80L, 29L, 12L, NA, 23L, 65L, NA, 130L, 
57L, NA, 46L, 109L, 69L, 66L, 10L, NA, 20L, 78L, 97L, 126L, 43L, 
58L, 120L, 107L, 66L, 76L, 23L, 69L, 64L, 98L, 96L, 135L, 41L, 
47L, 127L, 97L, 74L, 26L, 65L, 20L, 27L, 92L, 98L, 124L, 55L, 
53L, 121L, 96L, NA, 27L, NA, 73L, NA, 85L, 106L, 135L, 54L, NA
)), class = "data.frame", row.names = c(NA, -443L))

Answer 1

我们可以使用const char*用列的replace替换NA元素

median

或者另一个选择是library(dplyr) data %>% group_by(v2) %>% mutate_at(vars(v3, v4), list(~ replace(., is.na(.), median(., na.rm = TRUE)))) # A tibble: 443 x 4 # Groups: v2 [12] # v1 v2 v3 v4 # <int> <int> <dbl> <dbl> # 1 1908 1 118 10 # 2 1908 2 138 33 # 3 1908 3 128 17 # 4 1908 4 22.5 60 # 5 1908 5 42.5 114 # 6 1908 6 72 124 # 7 1908 7 82 44 # 8 1908 8 70 134 # 9 1908 9 58 121 #10 1908 10 42 117 # … with 433 more rows中的na.aggregate

zoo

默认情况下，library(zoo) data %>% group_by(v2) %>% mutate_at(vars(v3, v4), na.aggregate)是FUN。如果我们想要mean，请指定median

FUN

用group_by中值或平均值替换缺失值

1 个答案: