Elegant method for drawing hourly bar chart from time-interval data?

Question

I have a list of timesheet entries that show a start and stop time. This is sitting in a MySQL database. I need to create bar charts based on this data with the 24 hours of the day along the bottom and the amount of man-hours worked for each hour of the day.

For example, if Alice worked a job from 15:30 to 19:30 and Bob worked from 12:15 to 17:00, the chart would look like this:

I have a WTFey solution right now that involves a spreadsheet going out to column DY or something like that. The needed resolution is 15-minute intervals.

I'm assuming this is something best done in the database then exported for chart creation. Let me know if I'm missing any details. Thanks.

Answer 1

Create a table with just time in it from midnight to midnight containing each minute of the day. In the data warehouse world we would call this a time dimension. Here's an example:

TIME_DIM
 -id
 -time_of_day
 -interval_15 
 -interval_30

an example of the data in the table would be

id   time_of_day    interval_15    interval_30
1    00:00          00:00          00:00
...
30   00:23          00:15          00:00
...
100  05:44          05:30          05:30

Then all you have to do is join your table to the time dimension and then group by interval_15. For example:

SELECT b.interval_15, count(*) 
FROM my_data_table a
INNER JOIN time_dim b ON a.time_field = b.time
WHERE a.date_field = now()
GROUP BY b.interval_15

Answer 2

I came up with a pseudocode solution, hope it helps.

create an array named timetable with 24 entries
initialise timetable to zero

for each user in SQLtable
  firsthour = user.firsthour
  lasthour = user.lasthour

  firstminutes = 4 - (rounded down integer(user.firstminutes/15))
  lastminutes = rounded down integer(user.lastminutes/15)

  timetable(firsthour) = timetable(firsthour) + firstminutes
  timetable(lasthour) = timetable(lasthour) + lastminutes

  for index=firsthour+1 to lasthour-1
    timetable(index) = timetable(index) + 4
  next index

next user

Now the timetable array holds the values you desire in 15 minute granularity, ie. a value of 4 = 1 hour, 5 = 1 hour 15 minutes, 14 = 3 hours 30 minutes.

Answer 3

Here's another pseudocode solution from a different angle; a bit more intensive because it does 96 queries for every 24hr period:

results = []
for time in range(0, 24, .25):
  amount = mysql("select count(*) from User_Activity_Table where time >= start_time and time <= end_time")
  results.append(amount)

Answer 4

How about this:

Use that "times" table, but with two columns, containing the 15-minute intervals. The from_times are the 15-minutely times, the to_times are a second before the next from_times. For example 12:30:00 to 12:44:59.

Now get your person work table, which I've called "activity" here, with start_time and end_time columns.

I added values for Alice and Bob as per the original question.

Here's the query from MySQL:

SELECT HOUR(times.from_time) AS 'TIME', count(*) / 4 AS 'HOURS'
FROM times
  JOIN activity
  ON times.from_time >= activity.start_time AND 
     times.to_time   <= activity.end_time
GROUP BY HOUR(times.from_time)
ORDER BY HOUR(times.from_time)

which gives me this:

TIME   HOURS
12     0.7500
13     1.0000
14     1.0000
15     1.5000
16     2.0000
17     1.0000
18     1.0000
19     0.7500

Looks about right...