Question

I have a very big data frame consisting of data like this:

  PENR     ANFDAT     ENDDAT
1 1110 1990-02-01 1998-10-29
2 1981 1998-02-19 1998-02-20
3 6317 1994-11-01 1999-06-30
4 2039 1998-12-01 1999-04-30

(get it from here)

df <- structure(list(PENR = c(1110L, 1981L, 6317L, 2039L), ANFDAT = structure(c(7336, 10276, 9070, 10561), class = "Date"), ENDDAT = structure(c(10528, 10277, 10772, 10711), class = "Date")), .Names = c("PENR", "ANFDAT", "ENDDAT"), row.names = c(1L, 2L, 3L, 4L), class = "data.frame")

ANFDAT stands for the beginning of a certain status, ENDDAT stends for the termination of this status

I want to display this data as a bar chart. Each bar should stand for a date. the height of each bar should represent the number of records (PENR) having the status on this date.

As the data frame is very big I look for an efficient way to achieve this.

[EDIT]

Seems like my question is misleading. Here’s what I try to do:

1. Generate a data frame for each date from min(df$ANFDAT) to max(df$ENDDAT). This can be done easily with

   df1 <- data.frame(DATE = seq(min(df$ANFDAT), max(df$ENDDAT), by = "day"))
   

2. For each row in df1$DATE, count the number of records in df that have ANFDAT >= DATE and ENDDAT <= DATE. Store the results in df1$RECORDS

3. Generate a barplot out of df1. This probably can be done like that (untested)

   df2 <- df1$RECORDS
   row.names(df2) <- df1$DATE
   barplot(df2$RECORDS)
   

My problem is to find an efficient way to do #2

Answer 1

You could use sapply over the unique dates to count the number of records.

The whole procedure:

DATE <- seq.int(min(df$ANFDAT), max(df$ENDDAT), by = "day")

df1 <- data.frame(RECORDS = sapply(DATE, function(x)
                                sum(df[["ANFDAT"]] <= x &  df[["ENDDAT"]] >= x)))

rownames(df1) <- DATE
barplot(df1$RECORDS)