我在表格R中有一个数据框：
y
1484337600 19.22819
1484341200 19.28906
1484344800 19.28228
1484348400 19.21669
1484352000 19.32759
1484355600 19.21833
1484359200 19.20626
1484362800 19.28737
1484366400 19.20651
1484370000 19.18424
它有时代和价值观。大纪元时间是此数据帧中的row.names。现在，我希望将此数据帧转换为ARIMA所理解的ts（）。所以，我现在想到的是;我应该转换此数据帧zoo或xts，以便将我的纪元时间转换为日期时间，然后将其转换为时间序列对象，由arima识别以进行预测。此外，我找到了一个链接https://stats.stackexchange.com/questions/64383/what-is-the-best-forecasting-model-for-time-series-data，其中这些数据也具有纪元时间，然后将它们转换为xts然后转换为ts对象。但在我的情况下，纪元时间是索引列或row.names，我怎么做我的情况？这里的频率是1小时。
输入是上面给出的数据帧
我想将其转换为：
y
2017-01-13 13:00:00 19.28906
2017-01-13 14:00:00 19.28228
2017-01-13 15:00:00 19.21669
2017-01-13 16:00:00 19.32759
2017-01-13 17:00:00 19.21833
2017-01-13 18:00:00 19.20626
2017-01-13 19:00:00 19.28737
2017-01-13 20:00:00 19.20651
2017-01-13 21:00:00 19.18424
2017-01-13 22:00:00 19.25841
2017-01-13 23:00:00 19.15664
2017-01-14 00:00:00 19.16237
其中index是基于上述纪元时间的正确时间戳。然后我会在这个数据帧上调用ts（）。
我尝试使用以下方法执行此操作：
xts_dataframe <- xts(x = dataframe$y, order.by = as.POSIXct(as.numeric(row.names(dataframe)), origin="1970-01-01"))
**编辑： 我发现xts是由arima建模得到的，它可以工作。现在我关注的是上面的代码行是能够理解我的数据是每小时数据还是我需要做类似下面的事情来让它理解它是每小时的数据：
attr(xts_dataframe, 'frequency') <- 24
**我有这个疑问，因为： 案例1：当我在数据框架上运行arima时：
xts_dataframe <- xts(x = dataframe$y, order.by = as.POSIXct(as.numeric(row.names(dataframe)), origin="1970-01-01"))
产生的痕迹是这样的：
ARIMA(0,1,0) : 1054.432
ARIMA(0,1,0) with drift : 1056.433
ARIMA(0,1,1) : -22.90881
ARIMA(0,1,1) with drift : -20.90774
ARIMA(0,1,2) : -21.05988
ARIMA(0,1,2) with drift : -19.05796
ARIMA(0,1,3) : -130.5101
ARIMA(0,1,3) with drift : -128.5064
ARIMA(0,1,4) : -667.7424
ARIMA(0,1,4) with drift : -665.7444
ARIMA(0,1,5) : -669.2482
最佳模型是（4,1,1），所产生的预测是这样的：
Point Forecast Lo 80 Hi 80 Lo 95 Hi 95
15552001 18.86242 18.58299 19.14186 18.43507 19.28978
15555601 18.79712 18.48352 19.11072 18.31751 19.27673
15559201 18.94144 18.60992 19.27296 18.43443 19.44845
15562801 18.85743 18.47365 19.24120 18.27049 19.44436
15566401 18.84522 18.45196 19.23847 18.24378 19.44665
15570001 18.91105 18.51072 19.31139 18.29879 19.52331
15573601 18.86002 18.44583 19.27420 18.22657 19.49346
15577201 18.86466 18.44761 19.28172 18.22683 19.50249
15580801 18.89258 18.47185 19.31331 18.24913 19.53603
15584401 18.86414 18.43775 19.29054 18.21202 19.51626
15588001 18.87189 18.44333 19.30045 18.21646 19.52732
15591601 18.88261 18.45075 19.31446 18.22214 19.54307
15595201 18.86782 18.43197 19.30367 18.20125 19.53440
15598801 18.87426 18.43585 19.31267 18.20377 19.54475
显然，纪元时间相隔一小时。 （但是这些时代产生的时间比初始数据集少得多;这些时代属于1970年，当我转换并看到它时）
案例2：当我在数据框架上运行arima时，如：
xts_dataframe <- xts(x = dataframe$y, order.by = as.POSIXct(as.numeric(row.names(dataframe)), origin="1970-01-01"))
attr(xts_dataframe, 'frequency') <- 24
产生的痕迹如下所示：
ARIMA(0,1,0) : 1054.44
ARIMA(0,1,0) with drift : 1056.442
ARIMA(0,1,0)(0,0,1)[24] : 602.822
ARIMA(0,1,0)(0,0,1)[24] with drift : 604.8247
ARIMA(0,1,0)(0,0,2)[24] : 383.8319
ARIMA(0,1,0)(0,0,2)[24] with drift : 385.8355
ARIMA(0,1,0)(1,0,0)[24] : 389.8581
ARIMA(0,1,0)(1,0,0)[24] with drift : 391.8609
ARIMA(0,1,0)(1,0,1)[24] : Inf
ARIMA(0,1,0)(1,0,1)[24] with drift : Inf
ARIMA(0,1,0)(1,0,2)[24] : Inf
ARIMA(0,1,0)(1,0,2)[24] with drift : Inf
ARIMA(0,1,0)(2,0,0)[24] : 134.4009
ARIMA(0,1,0)(2,0,0)[24] with drift : 136.4046
ARIMA(0,1,0)(2,0,1)[24] : Inf
ARIMA(0,1,0)(2,0,1)[24] with drift : Inf
ARIMA(0,1,0)(2,0,2)[24] : Inf
ARIMA(0,1,0)(2,0,2)[24] with drift : Inf
ARIMA(0,1,1) : -22.90074
ARIMA(0,1,1) with drift : -20.89968
ARIMA(0,1,1)(0,0,1)[24] : -280.0375
ARIMA(0,1,1)(0,0,1)[24] with drift : -278.0347
ARIMA(0,1,1)(0,0,2)[24] : -422.9901
ARIMA(0,1,1)(0,0,2)[24] with drift : -420.9862
ARIMA(0,1,1)(1,0,0)[24] : -373.8774
ARIMA(0,1,1)(1,0,0)[24] with drift : -371.8742
ARIMA(0,1,1)(1,0,1)[24] : Inf
ARIMA(0,1,1)(1,0,1)[24] with drift : Inf
ARIMA(0,1,1)(1,0,2)[24] : Inf
ARIMA(0,1,1)(1,0,2)[24] with drift : Inf
ARIMA(0,1,1)(2,0,0)[24] : -551.472
ARIMA(0,1,1)(2,0,0)[24] with drift : -549.4676
ARIMA(0,1,1)(2,0,1)[24] : Inf
ARIMA(0,1,1)(2,0,1)[24] with drift : Inf
ARIMA(0,1,1)(2,0,2)[24] : Inf
ARIMA(0,1,1)(2,0,2)[24] with drift : Inf
ARIMA(0,1,2) : -21.0519
ARIMA(0,1,2) with drift : -19.04998
ARIMA(0,1,2)(0,0,1)[24] : -280.6503
ARIMA(0,1,2)(0,0,1)[24] with drift : -278.6467
ARIMA(0,1,2)(0,0,2)[24] : -423.2597
最佳模型又是（4,1,1），预测产生如下：
Point Forecast Lo 80 Hi 80 Lo 95 Hi 95
181.0000 18.86242 18.58299 19.14186 18.43507 19.28978
181.0417 18.79712 18.48352 19.11072 18.31751 19.27673
181.0833 18.94144 18.60992 19.27296 18.43443 19.44845
181.1250 18.85743 18.47365 19.24120 18.27049 19.44436
181.1667 18.84522 18.45196 19.23847 18.24378 19.44665
181.2083 18.91105 18.51072 19.31139 18.29879 19.52331
181.2500 18.86002 18.44583 19.27420 18.22657 19.49346
181.2917 18.86466 18.44761 19.28172 18.22683 19.50249
181.3333 18.89258 18.47185 19.31331 18.24913 19.53603
181.3750 18.86414 18.43775 19.29054 18.21202 19.51626
181.4167 18.87189 18.44333 19.30045 18.21646 19.52732
181.4583 18.88261 18.45075 19.31446 18.22214 19.54307
181.5000 18.86782 18.43197 19.30367 18.20125 19.53440
181.5417 18.87426 18.43585 19.31267 18.20377 19.54475
181.5833 18.87769 18.43584 19.31955 18.20193 19.55346
181.6250 18.87046 18.42500 19.31592 18.18919 19.55174
181.6667 18.87480 18.42638 19.32322 18.18900 19.56060
181.7083 18.87544 18.42353 19.32735 18.18430 19.56658
181.7500 18.87212 18.41677 19.32746 18.17573 19.56851
181.7917 18.87472 18.41623 19.33321 18.17352 19.57592
181.8333 18.87448 18.41257 19.33638 18.16806 19.58090
181.8750 18.87306 18.40784 19.33827 18.16157 19.58454
181.9167 18.87450 18.40610 19.34290 18.15814 19.59085
181.9583 18.87410 18.40239 19.34581 18.15268 19.59551
182.0000 18.87355 18.39862 19.34848 18.14721 19.59989
在这种情况下，如何转换日期以确保这些是未来的时间戳？
请确认哪种方式正确，但我从两者都得到相同的结果。