Earn the coveted Fabric Analytics Engineer certification. 100% off your exam for a limited time only!
Hi!
Has anyone tried to figure out how to calculate Erlang C formula with DAX? I am trying to calculate the number of agents needed for a specific level of service for a call centre.
I have followed this tutorial "Creating a Call Center staffing tool in Power PIvot" by Josh Miller. It was very helpful but Power BI is showing computing limitations, specifically is given me these errors:
“MdxScript(Model) (6,5) Calculation error in measure 'Table'[TA_INTERACTIONS]’: An argument of function FACT has the wrong data type or the result is too large or too small."
“MdxScript(Model) (25, 75) Calculation error in measure 'Table'[TA_INTERACTIONS]’: An argument of function POWER has the wrong data type or the result is too large or too small."
Calculations are based on Poisson and Cumulative Poisson formula, which contains a POWER() and FACT() functions:
a) Poisson:
PoissonX =
DIVIDE (
EXP ( - [IntensidadTraf] )
* POWER ( [IntensidadTraf]; SUM ( AgentsX[AgentsX] ) );
FACT ( SUM ( AgentsX[AgentsX] ) );
0
)
b) Cumulative Poisson
PoissonC_X =
SUMX (
FILTER ( AgentsZ; AgentsZ[AgentsZ] < SUM ( AgentsX[AgentsX] ) );
DIVIDE (
EXP ( - [IntensidadTraf] ) * POWER ( [IntensidadTraf]; AgentsZ[AgentsZ] );
FACT ( AgentsZ[AgentsZ] );
0
)
)
Where ‘AgentsX [AgentsX] ’ and ‘AgentsZ [AgentsZ] ’ are tables containing only one column with the total number of agents, increasing the value 1 by 1.
The model works for less than 150 agents but breaks at 151 agents or greater.
Any ideas on how to solve it?
Thanks!
I also found this helps too!
https://forum.enterprisedna.co/t/analysing-call-center-data/3146/2
How long of intervals were you using to figure (A) or Call intensity? Generally, something around 15 minutes - 30 minutes should work. There are some problems with the Erlang formula, in that the accuracy was only meant for ~143 agents. Anything above that, and it begins to break down.
I found this site fairly helpful on problemsolving and and overall helpful resource.
Sorry I wasn't able to find a solution to what to do if you have more agents than the Erlang calculations can handle, and I wish you well.
Hello Judy,
Have you resolved the Erlang - C logic issues or found any alternative solution?
Unfortunately Not yet
Hi @Anonymous ,
Actually, it mean you are use too large number to do 'fact' calculation and the result number is over int type limit.(I test on my side and I it can calculate with numbers less than 171)
Regards,
Xiaoxin Sheng
Thanks @v-shex-msft !
Though i need a much higher number to figure out my calculations. I will keep on searching an alternative.
Regards.
@Anonymous
I am trying to calculate call volume using Erlang C and Power BI variables. I am getting an error similiar to what you had. Were you able to find a way to forecast call volume in Power BI using Erlang?
User | Count |
---|---|
140 | |
113 | |
104 | |
76 | |
63 |
User | Count |
---|---|
135 | |
126 | |
110 | |
70 | |
61 |