Re: Commercial Alarm - help

[Nomen Nescio <nobody@xxxxxxxxx>]

Bass said:

>But cutting the number of false dispatches will have the affect of
>reducing the *precentage* of all
>alarms which are false.

You would think so, but it turns out the effect is insignificant.   I had
thought Jackcsg was just playing troll, enjoying an argument about whether
the sky is really blue or not.  But since you don't seem to get the math
either, others probably don't either.  Let's work through a realistic
scenario and look at the numbers.

Suppose a typical small alarm company has 500 accounts, all of which are
commercial.  Suppose that company has an average of one false alarm per
system per year, for a total 0f 500 false alarms.  Suppose that the company
also has the typical 99% false rate.  That means there are 505 total
dispatches, 500 false and 5 actual burglaries.

The NBFAA says that alarm companies should strive for a false alarm factor
of 0.50 for commercial accounts.  If this alarm company achieves that goal,
it will have 250 false alarms from its 500 accounts.  Presumably, the
number of actual burglaries will remain at 5.  There are 255 total
dispatches, 250 false and 5 actual burglaries, and that is 98% false.  A
50% reduction in the number of false alarms reduced the percentage by 1%.

Suppose this company does an even better job, and reduces its false alarms
down to 100.  That means each system, on average, causes only one false
alarm every five years.  Not bad for commercial systems.  But if there are
100 false alarms and 5 actual burglaries, the false alarm percentage is
95%.  An 80% reduction in the number of false alarms cut the percentage by
4%.

Now, suppose the number of actual burglaries drops from 5 to 3.  That
doesn't affect the number of false alarms, of course, but look what happens
to the percentages:  100 false alarms out of 103 total dispatches is 97%
false.

Suppose the number of actual burglaries increases from 5 to 10.  100 false
alarms out of 110 total dispatches is 91% false.

Strange as it may seem, a reduction in the number of burglaries increases
the false alarm rate, and an increase in the number of burglaries reduces
the false alarm rate.  Of course I am not advocating an increase in the
number of burglaries as a way to reduce the false alarm rate.  I'm just
pointing out the silliness of measuring alarm system reliability by using a
statistic that involves the number of burglaries.

>> Anywhere there is a large number of alarm systems and a relatively small
>> number of burglaries, the false alarm percentage will be extremely hig...
>
>That's patently untrue.  The fact is that police in the inner cities,
>where burglary rates are highest, usually experience the highest
>numbers (and percentages) of false alarms.

Just the opposite is true.  See the examples above.  Varying the number of
burglaries from 3 to 10 has a much greater effect on the false alarm
percentage than reducing the number of false alarms from 500 to 100.

Suppose our fictional alarm company knocks itself out, and cuts false
alarms all the way down to 50 per year (down from 500), for a false alarm
rate of 0.1 false alarms per system, per year.  Now, look at the false
alarm percentage for different numbers of actual burglaries:

50 false alarms...0 burglaries...false alarm percentage = 100%

50 false alarms...1 burglary...false alarm percentage = 98%

50 false alarms...2 burglaries...false alarm percentage = 96%

50 false alarms...3 burglaries...false alarm percentage = 94%

50 false alarms...4 burglaries...false alarm percentage = 93%

50 false alarms...5 burglaries...false alarm percentage = 91%

50 false alarms...50 burglaries...false alarm percentage = 50%

See what crap these percentages are?

- badenov