binning a vector in R

Question

[Assa Yeroslaviz]

Hi,

I have a data.frame of 17.7M rows, which I would like to plot as a heatmap (heatmap.2).

I would first like to create a table to see how many elements I have between (0,1], (1,2], (2,3]... ,(9,10]. I know that I have ~3500 elements without 'NA' and ~ 700 below 1. How can I do a complete overview of the data? Can I do it with table?

I would than like to bin the same data into specific (factorial?) values, so that I can use my own colors legend for each range.

This is my data structure:

> head(filtered1000)

      start     start.1     Value
1     876001    1403001     9.910803
2     3079001   3081001     9.891197
3     834001    836001      9.813543
4     1239001   1241001     9.794319
5     2777001   2780001     9.775171
6     1712001   1714001     9.727626

I know it is probably possible to manually adjust the cut command into the right breaks, but is there a more "subtle", automated way of calculating the number of each elements in each category and apply them to specific bins so that I will be able to set a certain color to a specific bin?

> breaks <- c(0,1,2,3,4,5,6,7,8,9,10)
> bins<-cut( filtered1000[,3], breaks )
summary(bins)

(0,1] (1,2] (2,3] (3,4] (4,5] (5,6] (6,7] (7,8] (8,9] (9,10]  NA's
2798  366   172   26    45    5     11    18    25    31      17762728

How can I than incorporate these breaks into my data.frame, so that I can plot (ggplot2) it with specific colors?

Thanks

Assa

Answer 1

[Devon Ryan]

It would be simpler to use findInterval().

[Assa Yeroslaviz]

Thanks. If I use findInterval(filtered1000[,3], seq(0:10)), I get the vector of 0-9 and a lot of NA, Can I than just plot them as if it is the third column?

How do I add it to my plot/data.frame, so that I can plot it with ggplot2?

[Devon Ryan]

Just add it to your data.frame using filtered1000$bins = findInterval(filtered1000[,3], seq(0:10)).

[Assa Yeroslaviz]

Oh man, sometimes it is sooooo easy.

Thanks