Need suggestion in doing functionality similar to What-if Analysis?

Question

Hi,
I have to develop a line chart with some requirements which is little bit tricky. So, was checking out if some one can help me out here. My scenario is explained below:
Suppose if you are a Oil drilling machinery supplier and you receive invoices for drilling machine order from your clients every month. Now last year because of the issue like BP oil spill in gulf of mexico, and this year Japan's Tsunami the client has not received many invoices from oil drilling companies which are working in those areas.
So my invoices numbers are like in Feb 2010 was 10000, in March 2010 was 11000, and in April 2010 was 11500, and after the incident in April 2010 the May 2010 values are 5000, June values are 5500, July is 5700, and August 2010 is 10300.
Same kind of deviated values happened after Japan Tsunami. ( i have just taken these examples to let you understand my problem clearly).
Now with the deviation in invoices after April 2010, and the business restoring to its normal sales after 4-5 months, let us assume that June 2011 invoices are 11900.
Now my user wants to see a line chart in which he will have last six months of data (Dec-2010 to May-2011). Now they want to have another series in the chart which is a prediction of the future three months values. Also they want an option where in they want to see if some disaster happens what could be the drop in their invoice numbers.
For the first prediction i think i can use Trend Analyzer and forecast it to three months by feeding the last 6 months of data and use the resultant values in the series. But for the second scenario what can i do to retrieve a deviation in values.
Please let me know if you need any other details here.

Answer 1

Nicolas,
I think my question was incomplete. Here i am not trying to see of how much deviation happened in my historical data but trying to find what could be the deviation in my future data. As i stated if the values for invoices in April 2010 to May 2010 dropped by 5000 because of a disaster, and after couple of months the business restored back to its regular invoices around 11000.
Now in future consider today is July 2011 and they got 11700 invoices already to be delivered for this month, they want to see a forecast or predicted value for August, September, October of 2011. For forecasting i used Trend Analyzer and feeded the last 12 months of data and i was able to get August, September, October values using the Linear Regression option in the component (the values were some were around 11767, 11821, 11853).
Now if some incident happens like i stated above like an oil spill for their major client and they stopped ordering the machinery how would the forecast values change. For this i understand that i have to feed a data as far as 4-5 yrs and see how deviation is happening because of disasters. But, can we do it manually in Xcelsius is what my question.
I went online and came to know of a tool called "Predictive Wrokbench" but the BI team Manager do not want to purchase the licence of a tool which is used only for only a single dashboard (which totally makes sense).