Here is a method of computing the count of deaths by date using awk and then printing the top 10

awk -F ‘\t’ ‘{if (NR>1 && length($4)==10) dates[$4]+=1} END {for (x in dates) printf “%s\t%s\n”,dates[x],x}’ deceased_person.tsv|sort -n|tail

I think the tail -11|head idiom was probably a remnant that I never cleaned up-maybe it had to do with a header row but truthfully I don’t remember – I should definitely clean this up Thanks for pointing it out.

why awk? cut is perfectly adequate, especially since tab is its default separator

$ cut -f1 deceased_person.tsv|sort|…

why the “|tail -11|head” idiom? if you want 10 items, just say “|tail”

s = sort relation by $1 desc;
h = limit s 10;

is equivalent to

g = group relation all;
h = foreach g { i = TOP(10,1,relation); generate flatten(i); }

but the second doesn’t require a sort (it’s all done in a single pass with a heap). this means no reduce step making it more composable by pig with other map tasks.

not sure if it’s doable in a generate block though… i’ll have to try it out sometime.

mat

hat tip to @squarecog for teaching me this one