readme.md

In ran-ran/zoocat: 'zoo' Objects with Column Attributes

title: "zoocat" output: md_document

output: html_document

variant: markdown_github

zoocat package is an extension package of the zoo package. The aim of this package is:

• Manipulate multivariate time serie conveniently.
• Convert between time series data and data frame.
• Represent multi-dimensional data by two-dimensional format, so we can focus on one dimension.

For using zoocat package, first load it:

library(zoocat)

Construct a zoocat object

Although column names can be used to identify columns, for "zoocat" class, a data frame is used to store information of columns, we call it column attribute table (cattr table). In the cattr table, each row represents a column of the core data in the object, and each column represents a column attribute. Using following codes to construct a zoocat object based on a matrix.

mat <- matrix(round(rnorm(24), 2), ncol = 4)
ctable <- data.frame(treatment = factor(rep(c('a', 'b'), 2), levels = c('a', 'b')), 
                     site = factor(rep(c('s1', 's2'), each = 2), levels = c('s1', 's2')))
zc <- zoocat(mat, order.by = 2011 : 2014, colattr = ctable, index.name = 'year')
print(zc)

#> A zoocat object with:
#> - [column attribute fields]: treatment, site
#> - [index variable]: year
#> - [data]:
#>       a_s1  b_s1  a_s2  b_s2
#> 2011  0.15  0.40 -0.45 -0.84
#> 2012  0.07 -1.65  0.03 -0.74
#> 2013 -0.16 -0.27  0.16  1.45
#> 2014 -1.10 -0.08 -0.75 -1.48

It can be seen that there are two column attributes: "treatment" and "site", and the index means "year". So, each data value corresponds to three underlying variables, i.e., treatment, site and year.

Set and get the cattr table

Using following codes to get and set cattr table.

cattr(zc)

#>   treatment site
#> 1         a   s1
#> 2         b   s1
#> 3         a   s2
#> 4         b   s2

zc2 <- zc
cattr(zc2) <- data.frame(cattr(zc), observer = 'Jack')
print(zc2)

#> A zoocat object with:
#> - [column attribute fields]: treatment, site, observer
#> - [index variable]: year
#> - [data]:
#>      a_s1_Jack b_s1_Jack a_s2_Jack b_s2_Jack
#> 2011      0.15      0.40     -0.45     -0.84
#> 2012      0.07     -1.65      0.03     -0.74
#> 2013     -0.16     -0.27      0.16      1.45
#> 2014     -1.10     -0.08     -0.75     -1.48

Merge by columns

Merging by columns is similar with "zoo" object except that the cattr tables need to be merged by rows. Note that NA is added if it is needed.

zc2 <- zc + 10
cattr(zc2) <- data.frame(site = 's3', added = rep(TRUE, 4))
zc.merge <- cbind(zc, zc2)
print(zc.merge)

#> A zoocat object with:
#> - [column attribute fields]: treatment, site, added
#> - [index variable]: year
#> - [data]:
#>      a_s1_NA b_s1_NA a_s2_NA b_s2_NA NA_s3_TRUE NA_s3_TRUE NA_s3_TRUE
#> 2011    0.15    0.40   -0.45   -0.84      10.15      10.40       9.55
#> 2012    0.07   -1.65    0.03   -0.74      10.07       8.35      10.03
#> 2013   -0.16   -0.27    0.16    1.45       9.84       9.73      10.16
#> 2014   -1.10   -0.08   -0.75   -1.48       8.90       9.92       9.25
#>      NA_s3_TRUE
#> 2011       9.16
#> 2012       9.26
#> 2013      11.45
#> 2014       8.52

cattr(zc.merge)

#>   treatment site added
#> 1         a   s1    NA
#> 2         b   s1    NA
#> 3         a   s2    NA
#> 4         b   s2    NA
#> 5      <NA>   s3  TRUE
#> 6      <NA>   s3  TRUE
#> 7      <NA>   s3  TRUE
#> 8      <NA>   s3  TRUE

Melt and cast

Similar with package reshape2, you can melt and cast between data frame and zoocat.

df.melt <- melt(zc)
head(df.melt)

#>   year treatment site value
#> 1 2011         a   s1  0.15
#> 2 2011         b   s1  0.40
#> 3 2011         a   s2 -0.45
#> 4 2011         b   s2 -0.84
#> 5 2012         a   s1  0.07
#> 6 2012         b   s1 -1.65

cast2zoocat(df.melt, index.var = 'year', value.var = 'value')

#> A zoocat object with:
#> - [column attribute fields]: treatment, site
#> - [index variable]: year
#> - [data]:
#>       a_s1  a_s2  b_s1  b_s2
#> 2011  0.15 -0.45  0.40 -0.84
#> 2012  0.07  0.03 -1.65 -0.74
#> 2013 -0.16  0.16 -0.27  1.45
#> 2014 -1.10 -0.75 -0.08 -1.48

Note that casting from a data frame is a convenient method to build a "zoocat" object.

Reset the index variable

The index variable can be reset to be a variable in the cattr table.

reset_index_var(zc, index.var = 'treatment')

#> A zoocat object with:
#> - [column attribute fields]: year, site
#> - [index variable]: treatment
#> - [data]:
#>   2011_s1 2011_s2 2012_s1 2012_s2 2013_s1 2013_s2 2014_s1 2014_s2
#> a    0.15   -0.45    0.07    0.03   -0.16    0.16   -1.10   -0.75
#> b    0.40   -0.84   -1.65   -0.74   -0.27    1.45   -0.08   -1.48

apply series functions

Functions can be applied for each column of the "zoocat" object, and the results will be binded with the cattr table.

apply_col(zc, FUN = mean)

#>   treatment site  output
#> 1         a   s1 -0.2600
#> 2         b   s1 -0.4000
#> 3         a   s2 -0.2525
#> 4         b   s2 -0.4025

apply_col(zc, FUN = function (x) {c(mean = mean(x), sd = sd(x))})

#>   treatment site    mean        sd
#> 1         a   s1 -0.2600 0.5752101
#> 2         b   s1 -0.4000 0.8797348
#> 3         a   s2 -0.2525 0.4228771
#> 4         b   s2 -0.4025 1.2777682

Another method is to apply a function for the whole core data, and the results can be binded with the cattr table or index.

apply_core(zc, FUN = colMeans, bind = 'cattr')

#>   treatment site data.ret
#> 1         a   s1  -0.2600
#> 2         b   s1  -0.4000
#> 3         a   s2  -0.2525
#> 4         b   s2  -0.4025

apply_core(zc, FUN = rowMeans, bind = 'index')

#>    2011    2012    2013    2014 
#> -0.1850 -0.5725  0.2950 -0.8525

apply_core(zc, FUN = function (x) {t(x*2)}, bind = c('cattr', 'index'))

#> A zoocat object with:
#> - [column attribute fields]: treatment, site
#> - [index variable]: index
#> - [data]:
#>       a_s1  b_s1  a_s2  b_s2
#> 2011  0.30  0.80 -0.90 -1.68
#> 2012  0.14 -3.30  0.06 -1.48
#> 2013 -0.32 -0.54  0.32  2.90
#> 2014 -2.20 -0.16 -1.50 -2.96

Filter columns

Function filter_col can be used to extract columns by specified conditions on the cattr table:

filter_col(zc, cond = treatment == 'a' & site == 's2')

#> A zoocat object with:
#> - [column attribute fields]: treatment, site
#> - [index variable]: year
#> - [data]:
#>       a_s2
#> 2011 -0.45
#> 2012  0.03
#> 2013  0.16
#> 2014 -0.75

Others

For "zoocat" class, all methods for the "zoo" class can be used. Addtionally, class "zoomly", based on "zoocat", is designed to manipulate monthly data.

ran-ran/zoocat documentation built on May 26, 2019, 10:56 p.m.