RoadMap GPS – Geneva Expert System

Published by RoadMap Technologies on March 1, 2014March 1, 2014

RoadMap Global Planning Solution

RoadMap’s Global Planning Solution provides four different enabling technologies that help organizations make accurate business plans for the future:

1. Geneva Expert System
2. Planning Workbooks
3. Data Warehouse
4. Cloud Deployment

Over the next few weeks, each of these technologies will be explained in detail on this blog – beginning with the Geneva Expert System.

The explosion of information technology created a huge challenge as well as an opportunity for business planning.  Product life cycles shortened.  New product launches occurred more rapidly. Point of sale and downstream distribution data gave planners a better view of consumer demand.

But forecasting was stuck back in the days of the IBM/360 mainframe computing.  An old dude wearing coke-bottle glasses, tethered to a mainframe computer terminal, mumbling about autocorrelations and ergodicity, working for hours at a time on a single product forecast. There had to be a better way… and we found it.

Sunspot Data by Year (NASA)

Instead of trying to forecast sales in the 21st century using the old forecasting methods which were used to forecast sunspot data starting from the year 1750, we figured that it would make more sense to start with business data.  Lots of it.

Through consulting engagements, we were able to assemble of huge database of product sales and forecasts from many of the leading global pharmaceutical companies, consumer electronics companies, and retailers.  But you can’t just surround a problem with data and wait for it to surrender; we had to go on the attack. But we didn’t use traditional statisticians – we needed people who weren’t trained forecasters.  We rounded up mechanical engineers, physicists, mathematicians and philosophers and turned them lose on this Big Data set with plenty of computing power.

What emerged after many iterations is the Geneva Expert System, which remains the most accurate business forecasting tool on the market today, compared to business intelligence tools and statistical packages.  Let’s do a quick case study to prove our point.

Remember last summer when it was actually hot?  After the first cookout in May, everyone had a great time, but the mosquitos were awful. Before the next cookout, people stocked up on Citronella candles.  Let’s look at weekly sales of Citronella candles – sales in June (~2500) are 10X sales in December.

Now let’s looks at the second line – The Geneva Expert System forecast.  Despite the erratic history, it automatically comes up with a model that you’d probably think was a reasonable starting point – then you could move on to the hundred or thousands of other products in your product line.

Citronella Candle – Weekly Sales History and Forecasts

But not so if you used the leading BI tool.  In that case, since the history ended around Labor Day when sales were declining, the forecast is for rapidly declining sales that end up with large negative numbers.

And the R statistical package using automated Box Jenkins? That statistical package forecasts around 880 in candle sales for each week.  Like a watch that is broken being right twice a day, that weekly forecast is correct only twice a year.

For RoadMap Global Planning, the Geneva Expert System gives planners an accurate and unbiased trend projection into the future.  It’s still up to the planner to assess what events are likely to “bend the trend” to arrive at an optimal forecast.