我有一个问题，我似乎无法解决。问题如下：我的数据看起来大概如此，它是一个data.table：
Code Date Marketvalue CommonEquity FiscalYearEnd
1: 13643N 2001-03-30 44.01 26761 2001-09-30
2: 13643N 2001-04-30 35.21 26761 2001-09-30
3: 13643N 2001-05-31 48.82 26761 2001-09-30
4: 13643N 2001-06-29 31.74 26761 2001-09-30
5: 13643N 2001-07-31 32.01 26761 2001-09-30
6: 13643N 2001-08-31 20.27 26761 2001-09-30
7: 13643N 2001-09-28 15.20 26761 2001-09-30
8: 13643N 2001-10-31 18.14 18618 2002-09-30
9: 13643N 2001-11-30 22.14 18618 2002-09-30
10: 13643N 2001-12-31 18.67 18618 2002-09-30
11: 13643N 2002-01-31 21.34 18618 2002-09-30
12: 13643N 2002-02-28 18.19 18618 2002-09-30
13: 13643N 2002-03-29 19.47 18618 2002-09-30
14: 13643N 2002-04-30 17.34 18618 2002-09-30
15: 13643N 2002-05-31 13.07 18618 2002-09-30
16: 13643N 2002-06-28 7.47 18618 2002-09-30
17: 13643N 2002-07-31 7.47 18618 2002-09-30
18: 13643N 2002-08-30 6.46 18618 2002-09-30
19: 13643N 2002-09-30 5.87 18618 2002-09-30
20: 13643N 2002-10-31 6.67 13728 2003-09-30
21: 13643N 2002-11-29 4.16 13728 2003-09-30
22: 13643N 2002-12-31 3.47 13728 2003-09-30
23: 13643N 2003-01-31 3.36 13728 2003-09-30
24: 13643N 2003-02-28 3.36 13728 2003-09-30
25: 13643N 2003-03-31 2.61 13728 2003-09-30
26: 13643N 2003-04-30 3.04 13728 2003-09-30
27: 13643N 2003-05-30 3.84 13728 2003-09-30
28: 13643N 2003-06-30 4.43 13728 2003-09-30
Code Date Marketvalue CommonEquity FiscalYearEnd
我想做什么：每年的每个六月我都希望将以t-1结尾的财政年度的普通股权与上一个12月的市场价值相匹配。现在获得去年12月的市值并不是那么难：我使用n = 6L的转移函数将12月市场价值转移到6月，但我如何用普通股来做？例如：对于2002年6月，该财政年度的财政年度结束时间是2002-09-30，但是我需要以t-1结尾的财政年度的普通股，在这种情况下是2001-09-30（值为26761，参见commonequity列）。任何人都可以帮我解决这个问题吗？我很遗憾没有任何我自己的代码来解决这个问题因为我根本不知道如何解决财政年度结束问题:(
提前致谢。
编辑：
我包括了预期的输出：
Code Date Marketvalue CommonEquity FiscalYearEnd BEMERatio
1: 13643N 2001-03-30 44.01 26761 2001-09-30 NA
2: 13643N 2001-04-30 35.21 26761 2001-09-30 NA
3: 13643N 2001-05-31 48.82 26761 2001-09-30 NA
4: 13643N 2001-06-29 31.74 26761 2001-09-30 NA
5: 13643N 2001-07-31 32.01 26761 2001-09-30 NA
6: 13643N 2001-08-31 20.27 26761 2001-09-30 NA
7: 13643N 2001-09-28 15.20 26761 2001-09-30 NA
8: 13643N 2001-10-31 18.14 18618 2002-09-30 NA
9: 13643N 2001-11-30 22.14 18618 2002-09-30 NA
10: 13643N 2001-12-31 18.67 18618 2002-09-30 NA
11: 13643N 2002-01-31 21.34 18618 2002-09-30 NA
12: 13643N 2002-02-28 18.19 18618 2002-09-30 NA
13: 13643N 2002-03-29 19.47 18618 2002-09-30 NA
14: 13643N 2002-04-30 17.34 18618 2002-09-30 NA
15: 13643N 2002-05-31 13.07 18618 2002-09-30 NA
16: 13643N 2002-06-28 7.47 18618 2002-09-30 1433.369
17: 13643N 2002-07-31 7.47 18618 2002-09-30 NA
18: 13643N 2002-08-30 6.46 18618 2002-09-30 NA
19: 13643N 2002-09-30 5.87 18618 2002-09-30 NA
20: 13643N 2002-10-31 6.67 13728 2003-09-30 NA
21: 13643N 2002-11-29 4.16 13728 2003-09-30 NA
22: 13643N 2002-12-31 3.47 13728 2003-09-30 NA
23: 13643N 2003-01-31 3.36 13728 2003-09-30 NA
24: 13643N 2003-02-28 3.36 13728 2003-09-30 NA
25: 13643N 2003-03-31 2.61 13728 2003-09-30 NA
26: 13643N 2003-04-30 3.04 13728 2003-09-30 NA
27: 13643N 2003-05-30 3.84 13728 2003-09-30 NA
28: 13643N 2003-06-30 4.43 13728 2003-09-30 5365.418
Code Date Marketvalue CommonEquity FiscalYearEnd BEMERatio
因为我们的数据表中没有2000年12月的市值，但我们在2002年6月和2003年6月都没有市值。我们没有对2006年6月的市场价值。基本上：每个6月：除以上一个日历年结束的会计年度的普通股按去年12月的市值计算。在这个例子中，2002年6月：26761（截至2001年的财政年度的普通股）除以18。67（2001年12月的市值）。
2003年6月：18618（截至2002年的财政年度的普通股）除以3.47（2002年12月的市值）
这是使用data.table的可能方法：
#calculate the prev fiscal year end date for self joining later
DT[!is.na(FiscalYearEnd),
PrevFiscalYearEnd := as.Date(vapply(FiscalYearEnd,
function(x) seq(x, by="-1 year", length.out=2L)[2L],
numeric(1L)),
origin="1970-01-01")]
ans <- DT[!is.na(FiscalYearEnd), BEMERatio :=
#get the CommonEquity for each fiscal year end
unique(DT[!is.na(FiscalYearEnd), .(Code, FiscalYearEnd, CommonEquity)])[
#join to find the previous fiscal year end common equity
DT[!is.na(FiscalYearEnd)], on=c("Code", "FiscalYearEnd"="PrevFiscalYearEnd")][,
#calculate ratio (need to make sure data is complete so that shifting by 6 months is robust)
CommonEquity / shift(Marketvalue, 6L)]
]
输出：
Code Date Marketvalue CommonEquity FiscalYearEnd PrevFiscalYearEnd BEMERatio
1: 13643N 2001-03-30 44.01 26761 2001-09-30 2000-09-30 NA
2: 13643N 2001-04-30 35.21 26761 2001-09-30 2000-09-30 NA
3: 13643N 2001-05-31 48.82 26761 2001-09-30 2000-09-30 NA
4: 13643N 2001-06-29 31.74 26761 2001-09-30 2000-09-30 NA
5: 13643N 2001-07-31 32.01 26761 2001-09-30 2000-09-30 NA
6: 13643N 2001-08-31 20.27 26761 2001-09-30 2000-09-30 NA
7: 13643N 2001-09-28 15.20 26761 2001-09-30 2000-09-30 NA
8: 13643N 2001-10-31 18.14 18618 2002-09-30 2001-09-30 760.0398
9: 13643N 2001-11-30 22.14 18618 2002-09-30 2001-09-30 548.1565
10: 13643N 2001-12-31 18.67 18618 2002-09-30 2001-09-30 843.1317
11: 13643N 2002-01-31 21.34 18618 2002-09-30 2001-09-30 836.0200
12: 13643N 2002-02-28 18.19 18618 2002-09-30 2001-09-30 1320.2269
13: 13643N 2002-03-29 19.47 18618 2002-09-30 2001-09-30 1760.5921
14: 13643N 2002-04-30 17.34 18618 2002-09-30 2001-09-30 1475.2481
15: 13643N 2002-05-31 13.07 18618 2002-09-30 2001-09-30 1208.7173
16: 13643N 2002-06-28 7.47 18618 2002-09-30 2001-09-30 1433.3690
17: 13643N 2002-07-31 7.47 18618 2002-09-30 2001-09-30 1254.0300
18: 13643N 2002-08-30 6.46 18618 2002-09-30 2001-09-30 1471.1930
19: 13643N 2002-09-30 5.87 18618 2002-09-30 2001-09-30 1374.4735
20: 13643N 2002-10-31 6.67 13728 2003-09-30 2002-09-30 1073.7024
21: 13643N 2002-11-29 4.16 13728 2003-09-30 2002-09-30 1424.4836
22: 13643N 2002-12-31 3.47 13728 2003-09-30 2002-09-30 2492.3695
23: 13643N 2003-01-31 3.36 13728 2003-09-30 2002-09-30 2492.3695
24: 13643N 2003-02-28 3.36 13728 2003-09-30 2002-09-30 2882.0433
25: 13643N 2003-03-31 2.61 13728 2003-09-30 2002-09-30 3171.7206
26: 13643N 2003-04-30 3.04 13728 2003-09-30 2002-09-30 2791.3043
27: 13643N 2003-05-30 3.84 13728 2003-09-30 2002-09-30 4475.4808
28: 13643N 2003-06-30 4.43 13728 2003-09-30 2002-09-30 5365.4179
Code Date Marketvalue CommonEquity FiscalYearEnd PrevFiscalYearEnd BEMERatio
如果你真的需要在6月以外的几个月做NA：
ans[month(Date)!=6L, BEMERatio := NA_real_]
输出：
Code Date Marketvalue CommonEquity FiscalYearEnd PrevFiscalYearEnd BEMERatio
1: 13643N 2001-03-30 44.01 26761 2001-09-30 2000-09-30 NA
2: 13643N 2001-04-30 35.21 26761 2001-09-30 2000-09-30 NA
3: 13643N 2001-05-31 48.82 26761 2001-09-30 2000-09-30 NA
4: 13643N 2001-06-29 31.74 26761 2001-09-30 2000-09-30 NA
5: 13643N 2001-07-31 32.01 26761 2001-09-30 2000-09-30 NA
6: 13643N 2001-08-31 20.27 26761 2001-09-30 2000-09-30 NA
7: 13643N 2001-09-28 15.20 26761 2001-09-30 2000-09-30 NA
8: 13643N 2001-10-31 18.14 18618 2002-09-30 2001-09-30 NA
9: 13643N 2001-11-30 22.14 18618 2002-09-30 2001-09-30 NA
10: 13643N 2001-12-31 18.67 18618 2002-09-30 2001-09-30 NA
11: 13643N 2002-01-31 21.34 18618 2002-09-30 2001-09-30 NA
12: 13643N 2002-02-28 18.19 18618 2002-09-30 2001-09-30 NA
13: 13643N 2002-03-29 19.47 18618 2002-09-30 2001-09-30 NA
14: 13643N 2002-04-30 17.34 18618 2002-09-30 2001-09-30 NA
15: 13643N 2002-05-31 13.07 18618 2002-09-30 2001-09-30 NA
16: 13643N 2002-06-28 7.47 18618 2002-09-30 2001-09-30 1433.369
17: 13643N 2002-07-31 7.47 18618 2002-09-30 2001-09-30 NA
18: 13643N 2002-08-30 6.46 18618 2002-09-30 2001-09-30 NA
19: 13643N 2002-09-30 5.87 18618 2002-09-30 2001-09-30 NA
20: 13643N 2002-10-31 6.67 13728 2003-09-30 2002-09-30 NA
21: 13643N 2002-11-29 4.16 13728 2003-09-30 2002-09-30 NA
22: 13643N 2002-12-31 3.47 13728 2003-09-30 2002-09-30 NA
23: 13643N 2003-01-31 3.36 13728 2003-09-30 2002-09-30 NA
24: 13643N 2003-02-28 3.36 13728 2003-09-30 2002-09-30 NA
25: 13643N 2003-03-31 2.61 13728 2003-09-30 2002-09-30 NA
26: 13643N 2003-04-30 3.04 13728 2003-09-30 2002-09-30 NA
27: 13643N 2003-05-30 3.84 13728 2003-09-30 2002-09-30 NA
28: 13643N 2003-06-30 4.43 13728 2003-09-30 2002-09-30 5365.418
Code Date Marketvalue CommonEquity FiscalYearEnd PrevFiscalYearEnd BEMERatio