This is output from a script I cooked up for comparing today’s visitor counts to yesterday’s. The idea is to see if the counts are in line with normal activity: Hourly counts that went low would mean something wrong with the web site. High counts, which happen occasionally, mean that someone gave Entrance a plug somewhere.
(This chart was made a little after 9 AM this morning, so the red bar for 9 is low representing an incomplete count)
To ensure that any 0 counts show up, I include a LEFT JOIN with a table containing the values 0-23. SKINNY makes the second set of bars 1/3 the width of the first set. Here’s the complete script:
-- hours 0 - 23
create temporary table hours_t
(h int, today int, yesterday int);
insert into hours_t(h) values
(0), (1), (2), (3), (4), (5), (6), (7), (8),
(9), (10), (11), (12), (13), (14), (15), (16),
(17), (18), (19), (20), (21), (22), (23);
create temporary table today_t
select h, count(distinct ip) c
from ACCESS
where day(dtime) = day(curdate())
group by h;
create temporary table yesterday_t
select h, count(distinct ip) c2
from ACCESS
where date(dtime) =
date(date_sub(curdate(), interval 1 day))
group by h;
plot
all axislabels, very light blue bar, light red bar
with
no sides
baroverlap 1
skinny
comment left "Number of Visitors"
title x "Hour of Day"
select hours_t.h, c2, c
from (hours_t LEFT JOIN today_t
on hours_t.h = today_t.h) LEFT JOIN
yesterday_t
on hours_t.h = yesterday_t.h;
With a similar technique, you could compare this months performance to last months, or days in this weeks to last weeks, etc.