Poisson and Contact Centre Forecasts

Topic Views - 2171

I am looking for more info on how to use poisson distribution to report on forecast accuracy. I have heard it mentioned a few times in conferences and am looking to find out more about it in practice. Like what is the formula, how is it used, what does the output mean etc


Call Centre Helper

Generally Poisson is not used to calculate forecast accuracy
Poisson tends to be used in contact centres as the underlying statistical method for the Erlang C formula which is used to work out staffing levels. The Erlang fomula uses poisson distribution.

In terms of Forecast accuracy the best way to do it is to use something like MAPE (Mean Absolute Percentage Error) measured on each interval - see this article

It is possible that you could then look at the distribution of forecast accuracy and see how many forecasts periods fall outside of one standard distribution. This can be a useful technique for spotting anomalies in data.

Thanks for this. ...
Thanks for this.

The problem I have is the interval numbers are very small. Some teams take as many as 40 calls a day and % differences look terrible when they are a handful of calls out on forecast. I may have misunderstood when to utilise Poisson distribution as I understood it to be ideal for those sort of scenarios. I may try the standard deviation approach instead.


Call Centre Helper

Forecasting intermittent demand
Certainly Poisson is good for forecasting intermittent demand.

It sounds like you are experiencing a high degree of volativity. What is your typical MAPE for the day (based on interval)?

Want to add a comment?

Not found what you were looking for?

1. Try searching through our site.
2. Still not got an answer?

Why not ask the Call Centre Helper Community? Click here to ask your question