Example 9.28: creating datasets from tables

R
There are often times when it is useful to create an individual level dataset from aggregated data (such as a table). While this can be done using the expand.table() function within the epitools package, it is also straightforward to do directly within R.

Imagine that instead of the individual level data, we had only the 2x2 table for the association between homeless status and gender within the HELP RCT:

> HELPrct = read.csv("http://www.math.smith.edu/r/data/help.csv")
> xtabs(~ homeless + female, data=HELPrct)
        female
homeless   0   1
       0 177  67
       1 169  40

We can use this information to create an analytic dataset using just the four rows of a new dataset:

> female = c(0, 1, 0, 1)
> homeless = c(1, 1, 0, 0)
> count = c(169, 40, 177, 67)
> ds=data.frame(cbind(female, homeless, count))
> ds
  female homeless count
1      0        1   169
2      1        1    40
3      0        0   177
4      1        0    67

Next we use the rep() function to generate a vector of indices to repeat. The index object repeats each row number count times.

> index = rep(seq_len(nrow(ds)), times=ds$count)
> newds = ds[index,]
> newds$count = NULL
> xtabs(~ homeless + female, data=newds)
        female
homeless   0   1
       0 177  67
       1 169  40

The resulting data set is identical to the summarized input data set.

SAS
Many SAS procedures offer a weight varname option (as a statement within the proc) which will duplicate each observation varname times. So, for example, we can make a data set such as that shown above, then use, e.g., proc freq to produce a table.

data ds;
female = 0; homeless = 1; count = 169; output;
female = 1; homeless = 1; count = 40; output;
female = 0; homeless = 0; count = 177; output;
female = 1; homeless = 0; count = 67; output;
run;

proc freq data = ds;
table homeless * female;
weight count;
run;
                homeless     female

                Frequency|
                Percent  |
                Row Pct  |
                Col Pct  |       0|       1|  Total
                ---------+--------+--------+
                       0 |    177 |     67 |    244
                         |  39.07 |  14.79 |  53.86
                         |  72.54 |  27.46 |
                         |  51.16 |  62.62 |
                ---------+--------+--------+
                       1 |    169 |     40 |    209
                         |  37.31 |   8.83 |  46.14
                         |  80.86 |  19.14 |
                         |  48.84 |  37.38 |
                ---------+--------+--------+
                Total         346      107      453
                            76.38    23.62   100.00

However, some procedures lack this option, and/or it may be difficult to arrange your data appropriately to take advantage of it. In such cases, it's useful to be able to expand the data manually, as we show for R above. We demonstrate this below, assuming the count variable can be constructed. The explicit output statement puts a line into the newds data set count times.

data newds;
set ds;
do i = 1 to count;
  output;
  end;
run;

proc freq data = newds;
table homeless * female;
run;
                homeless     female

                Frequency|
                Percent  |
                Row Pct  |
                Col Pct  |       0|       1|  Total
                ---------+--------+--------+
                       0 |    177 |     67 |    244
                         |  39.07 |  14.79 |  53.86
                         |  72.54 |  27.46 |
                         |  51.16 |  62.62 |
                ---------+--------+--------+
                       1 |    169 |     40 |    209
                         |  37.31 |   8.83 |  46.14
                         |  80.86 |  19.14 |
                         |  48.84 |  37.38 |
                ---------+--------+--------+
                Total         346      107      453
                            76.38    23.62   100.00

An unrelated note about aggregatorsWe love aggregators! Aggregators collect blogs that have similar coverage for the convenience of readers, and for blog authors they offer a way to reach new audiences. SAS and R is aggregated by R-bloggers and PROC-X with our permission, and by at least 2 other aggregating services which have never contacted us. If you read this on an aggregator that does not credit the blogs it incorporates, please come visit us at SAS and R. We answer comments there and offer direct subscriptions if you like our content. In addition, no one is allowed to profit by this work under our license; if you see advertisements on this page, the aggregator is violating the terms by which we publish our work.