R-缺少数据的时间序列的插值

Question

I have a database with time data, but for some timestamps there is no data available (NA in database). I would like to do an interpolation for these values.

Dataset:

 structure(list(timestamp = structure(1:7, .Label = c("21/01/2012 18:41", 
+ "21/01/2012 18:46", "21/01/2012 18:51", "21/01/2012 18:56", "21/01/2012 19:01", 
+ "21/01/2012 19:06", "21/01/2012 19:11"), class = "factor"), humid = c(47.7, 
+ 44.5, NA, 42.5, 42.5, NA, 41.6), temp = c(14.12, 15.37, NA, 16.17, 
+ 16.31, NA, 16.51)), .Names = c("timestamp", "humid", "temp"), class = "data.frame", row.names = c(NA, 
+ -7L))

Which looks like this:

         timestamp              humid               temp
1 21/01/2012 18:41 47.700000000000003 14.119999999999999
2 21/01/2012 18:46 44.500000000000000 15.369999999999999
3 21/01/2012 18:51                 NA                 NA
4 21/01/2012 18:56 42.500000000000000 16.170000000000002
5 21/01/2012 19:01 42.500000000000000 16.309999999999999
6 21/01/2012 19:06                 NA                 NA
7 21/01/2012 19:11 41.600000000000001 16.510000000000002

I already tried the option A:

library(zoo)
Mz <- zoo(TEST)
index(Mz) <- Mz[,1]
Mz_approx <- na.approx(Mz, x=Mz$timestamp)

But this results in following errors:

Error in approx(x[!na], y[!na], xout, ...) : 
  need at least two non-NA values to interpolate
In addition: Warning messages:
1: In na.approx.default(object, x = x, xout = xout, na.rm = FALSE,  :
  NAs introduced by coercion
2: In na.approx.default(object, x = x, xout = xout, na.rm = FALSE,  :
  NAs introduced by coercion
3: In xy.coords(x, y) : NAs introduced by coercion

I also tried option B:

library(zoo)
Mz <- zoo(TEST)
Mz_approx <- na.approx(Mz)

But this results in the following errors:

Error in approx(x[!na], y[!na], xout, ...) : 
  need at least two non-NA values to interpolate
In addition: Warning message:
In xy.coords(x, y) : NAs introduced by coercion

What is the best way to overcome these errors and to use the function na.approx correctly?

Answer 1

read.zoo will convert it to zoo handling the index properly and then na.approx can be used. There are several vignettes (pdf manuals) that come with zoo including an entire manual just devoted to read.zoo examples and there are many examples in the zoo help files that you can go through.

library(zoo)

z <- read.zoo(TEST, tz = "", format = "%d/%m/%Y %H:%M")
na.approx(z)

giving:

                    humid  temp
2012-01-21 18:41:00 47.70 14.12
2012-01-21 18:46:00 44.50 15.37
2012-01-21 18:51:00 43.50 15.77
2012-01-21 18:56:00 42.50 16.17
2012-01-21 19:01:00 42.50 16.31
2012-01-21 19:06:00 42.05 16.41
2012-01-21 19:11:00 41.60 16.51