Lucent Technologies CentreVu Call Management System Release 3 Version 5 Manual

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-9
Seasonal Trending: 
1. Find the Seasonal Trending Ratio (STR).
a. Multiply the Number of Calls Carried (NCC) for seasonal days 
S1 through S4 by the assigned data weights.
The NCC for a seasonal day includes the NCC for the whole 
day, not just one intrahour interval.
b. Add the weighted data together and divide that sum by the sum 
of the individual weights.
c. Divide the NCC for season day S0 by the quotient found in 
Step b.
The equation for STR is as follows:
where 
W1 through W4 are the corresponding data weights.
2. Find the Weighted Average of Calls Carried (WACC) of the current 
data.
3. Divide the Change Factor (CF) by 100.
4. Multiply the factors found in the previous three steps to get the FCC 
for each intrahour interval.
STR
NCCS0
W1NCCS1´ ()W2NCCS2´ ()W3NCCS3´ ()W4NCCS´ ( +++
W1W2W3W4+++---------------------------------------------------------------------------------------------------------------------------------------------------------
-----------------------------------------------------------------------------------------------------------------------------------------------------------=
WACCW1xNCCC1()W2xNCCC2()W3xNCCC3()W4xNCCC4() +++
W1W2W3W4+++ -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- =
FCC STR WACCCF100----------´ ´= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-10
Current Trending: 
1. Find the projected number of calls in each intrahour interval of the 
forecast date using a weighted quadratic least squares curve fit.
The weighted quadratic least squares curve fit is an algorithm that 
fits a curve to the current data. The algorithm uses the fitted curve 
as a trend, and projects the number of calls based on this trend. The 
algorithm works in the following manner:
lIf four data points are given (all four points have weights 
greater than 0), the algorithm derives a parabolic curve that fits 
the four points with the least error. The algorithm then uses this 
curve to project the number of calls for the forecast date.
lIf three data points are given (one point has a weight of 0), the 
algorithm derives a straight line that runs through the three 
points with the least error. The algorithm uses the slope of this 
line to project the number of calls for the forecast date.
lIf two data points are given, the algorithm returns the weighted 
average of the two points as the projected number of calls.
lIf one data point is given, that value is returned as the projected 
number of calls.
2. Divide the Change Factor (CF) by 100.
3. Multiply the factors found in the previous two steps to get the FCC 
for each intrahour interval.
FCC projectedcallsCF100----------´= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-11
Expected Calls:  
1. Find the Weighted Average of Calls Carried (WACC) of the current 
data.
2. Divide the WACC for each interval by the sum of WACC for all 
intervals.
3. Multiply the quotient found in Step 2 by the Expected Calls (EC) to 
get the FCC for each intrahour interval.
WACCW1xNCCC1()W2xNCCC2()W3xNCCC3()W4xNCCC4() +++
W1W2W3W4+++ -------------------------------------------------------------------------------------------------------------------------------------------------------------------------- =
FCCWACC each interval
WACC all intervals ------------------------------------------------------EC´= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-12
Algorithm for FCC 
Intraday
6
This algorithm computes the FCC for each intrahour interval in Intraday 
reports. The algorithm uses the NCC from intervals that have already 
passed and the FCC in the Current Day report to obtain a ratio of actual 
calls to forecast calls. Then, using this ratio, the algorithm scales the FCC 
for the remaining intervals in the day. The number of agents required to 
handle the new call volume is recalculated.
1. Sum together the NCC for all specified passed intervals.
2. Sum together the FCC for all specified passed intervals.
3. Divide the sum found in Step 1 by the sum found in Step 2.
4. Multiply the FCC for the remaining intervals in the Current Day 
report by the quotient found in Step 3.
5. The equation for FCC is as follows:
F C C FCC remaining intervalNCC passed intervals
FCC passed intervals ----------------------------------------------------------´= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-13
Algorithm for FCC 
Special Days
6
This algorithm computes the FCC for each intrahour interval in Special 
Day reports. The algorithm computes the FCC using historical data 
collected from the previous special day. There is no trending; however, 
you may or may not specify Expected Calls which results in the following 
two algorithms:
No Expected Calls: 
1. Find the NCC in each intrahour interval of the previous special day.
2. Divide the Change Factor (CF) by 100.
3. Multiply the factors found in the previous two steps to get the FCC 
for each intrahour interval.
Expected Calls: 
This algorithm distributes the Expected Calls into intrahour intervals in a 
pattern similar to the NCC of the previous special day.
1. Divide the NCC in each interval of the previous special day by the 
sum of NCC in all intervals of that day.
2. Multiply the quotient found in Step 2 by the Expected Calls (EC) to 
get the FCC for each intrahour interval.
F C C NCC each interval Prev Spec DayCF100----------´=
FCCNCC each interval Prev Spec Day
NCC all intervals Prev Spec Day ------------------------------------------------------------------------------------------EC´= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-14
Algorithm for 
Number of 
Agents Required
6
This algorithm computes the Number of Agents Required for each 
intrahour interval in the following reports:
lLongterm
lFinancial
lCurrent Day
lIntraday
lSpecial Day
lHypothetical
lHypothetical Financial.
Use the following procedure:
1. Convert FCC to traffic load in Erlangs.
2. Set the number of agents to the load + 1.
3. Compute the initial probability that all agents are busy.
4. Compute subsequent values for the probability with the following 
formula:
where 
P = probability all agents are busy, n = number of agents, 
a = load in Erlangs.
5. Using the above probability, the number of agents, and the given 
load, compute a new set of objectives which correspond to the 
objectives in the Call Handling Profile.
6. If any of the new objectives exceeds or falls below the objectives in 
the Call Handling Profile, end the search. Return the number of 
agents as the answer. Otherwise, add 1 to the number of agents, 
and repeat Steps 4 through 6.
Pa
n
n1– ()!na– ()-------------------------------------
1a
k
k!-----a
n
n1– ()!na– ()------------------------------------- +
k1= n1–
å+ ------------------------------------------------------------------------ =
P
n1
1n1
–aPn1––
n1–a– ()Pn1–-----------------------------------------xna–
a ------------ + --------------------------------------------------------------------- = 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Call Volume/Agents Forecast Reports6-15
Algorithm for 
Estimated Margin
6
This algorithm computes the Estimated Margin for each intrahour interval 
in Financial reports. The Estimated Margin is the difference between call 
revenue and call costs.
1. Multiply the Number of Agents Required (NAR) by the cost of each 
agent.
2. Multiply the Forecast Calls Carried (FCC) by the cost of each call.
3. Add together the products found in Steps 1 and 2.
4. Multiply the FCC by the revenue from each call.
5. Subtract the agent and call costs from the revenue to get the 
Estimated Margin (EM).
EM FCC call revenue´()N A R agent cost´()FCC call cost´()+[] –= 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Requirement Reports6-16
Requirement Reports6
This section includes all of the algorithms used in Requirement Reports:
l“Algorithm for Agent Positions Required” 
l“Algorithm for Trunks Required” .
Algorithm for 
Agent Positions 
Required
6
This algorithm computes how many calls a given number of agents can 
handle per intrahour interval. The number of agents can be a number or a 
range of numbers. The results of the algorithm are within the call 
handling objectives you specify in the report.
1. Determine the number of agents for which the calls carried is to be 
computed.
2. For each number of agents, compute the load.
The following search calculates the load using an upper and lower 
bound:
a. Set the lower bound for the load to 0.
b. Set the upper bound for the load to the number of agents.
c. Compute the midpoint.
d. Compute the probability that all agents are busy using the 
midpoint as the load.
where 
P = probability all agents are busy, n = number of agents, 
and 
a = load in Erlangs (midpoint).
e. Compute a new set of call handling objectives using the above 
probability.
Midpointlower bound upperbound+
2 ----------------------------------------------------------------------- =
Pa
n
n1– ()!na– ()-------------------------------------
1a
k
k!-----
k1
n1–
åan
n1– ()!na– ()------------------------------------- ++ ------------------------------------------------------------------------ = 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Requirement Reports6-17
f. If any of the new objectives exceeds or falls below the 
objectives in the Call Handling Profile, set upper bound = 
midpoint. Otherwise, set lower bound = midpoint.
g. Compute the new midpoint.
h. If the new midpoint changed less than 0.001 from the previous 
midpoint, end the search. Return the midpoint as the load. 
Otherwise, repeat Steps d through h using the new midpoint.
3. Convert the load into calls carried.
When you enter a range of numbers, you can set graduations to yes 
or no. With no graduations, the algorithm does its computations for 
every number in the range. With graduations, the algorithm starts its 
computations with the closest number in the following set and 
proceeds as follows: 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 12, 14, 16, 18, 20, 
25, 30, 35, 40, 45, 50, 60, 70, 80, 90, 100...(continuing by tens).
Algorithm for 
Trunks Required
6
This algorithm computes how many calls a given number of trunks can 
handle per intrahour interval. The number of trunks can be a number or a 
range of numbers. The results of the algorithm are within the trunk 
blocking probability and the average trunk holding time you specify in the 
report.
1. Determine the number of trunks for which the calls carried is to be 
computed.
2. For each number of trunks, compute the load.
The following search computes the load using an upper and lower 
bound:
a. Set the lower bound for the load to 0.
b. Set the upper bound for the load to the number of trunks, and, 
while the blocking probability produced by the number of trunks 
and the high load is ³ blocking objective, double the upper 
bound.
c. Compute the midpoint.
d. Calculate the blocking probability using the midpoint as the 
load.
Midpointlower bound upperbound+
2 ----------------------------------------------------------------------- = 
						

							  How the Forecast System Generates Data CentreVu CMS R3V5 Forecast  585-215-825
Requirement Reports6-18
where P = blocking probability, n = number of trunks, 
a = load in Erlangs (midpoint).
e. If 
P £ blocking objective, set the lower bound = midpoint. 
Otherwise, set the upper bound = midpoint.
f. If the midpoint is close enough, return the midpoint as the 
desired load. Otherwise, repeat Steps 
c through f.
This algorithm handles graduations the same way as the Agent Positions 
Required algorithm.
Pa
n
n!------
1a
k
k!-----
k1=n
å+
----------------------------- =