ISIS Solutions, Inc. distinguishes itself by providing Forecasting and Predictive Analytics that looks into the future as compared to Business Intelligence tools that are focused on reporting on the past.  ISIS Discovery & Predictive Analytic software applies sophisticated mathematics to enterprise data to show the meaning of the data and predict the future.  Looking forward makes managers proactive rather than reactive as is the result of historical reporting.

The figure below shows a complete information landscape and the progression of manipulating data from reporting to becoming a Predictive Business.  Data can be manipulated in three dimensions of Organization, Time, and Mathematical Sophistication.  Organization refers to the detail of the data (such as, detail products up to the corporate level).  Time covers the past, present and future.  Mathematical Sophistication ranges from none (just reporting data) to predictive statistical probability calculations.

Reporting & Analysis

Business Intelligence tools cover the first two segments of data manipulation to "see" and "compare" data.  Comparison is what BI vendors refer to as "analysis" where, for example, a salesman's performance is being compared this year to the same period last year and a variance in dollars and percent is calculated.  These static reports provide historical results and offer access to data.

Analytics

Following reporting and analysis is Analytics where mathematics is applied to data.  Here we go from reporting on data to learning what data means.  For example, the graph below shows the cumulative Units sold, Revenue and Gross Margin for a particular product.  What is this graph telling us?  Relatively nothing!

We see parallel curves for Units and Revenue and a straight line for the Gross Margin but how is this information used?  Further, is business for this product good and getting better?  What does it mean that the Gross Margin is diverging from the Gross Revenue?  What if this is the graph for one product, what happens if we have 1,000 products or 10,000 products?  Are we supposed to visually inspect each graph to divine an answer?

Now let's take the same data in the graph above but do a series of analytics.  First, let's use Units as an "Activity Driver" to apply against Gross Margin to calculate efficiency of the product to generate Gross Profit per Unit Sold, shown as the Blue curve in the graph below.

Now the data has meaning to reveal the product's lifecycle.  In the first two months, the product is in its "Early Stage" as the efficiency to generate Gross Margin per Unit Sold is increasing.  In months three and four the Gross Margin per Unit Sold efficiency has peaked and the product is in its "Mature Stage".  By month five the Gross Profit per Unit Sold is strongly declining and is in its "Late Stage".

Next, let's calculate the Slope.  Calculating the Slope of the Gross Margin per Unit Sold at any point in time reveals the leading indicator of the future of the trend.  In this example, had the Slope been calculated at month two, the Purple line in the graph above, it would have been seen that the trend for months one and two was about to break down and move the product to a Mature Stage from Early Stage.  At that point proactive decisions could be made about investments in the product to keep it in the Early stage or if its over-all contribution to profitability is less significant then introduce lower mark-downs early to expedite flushing it through the channels and minimize the impact on profitability.

Predictive Forecasting™

Forecasting properly defined is the application of mathematical formula to historical data to calculate a value at a future point in time.  Since not all business or all items in a business behave the same the accuracy of the forecast is dependent on having a formula that can describe the characteristic of the item to be forecast.  This finding was validated by research at the University of Pennsylvania by Fred Collopy and J. Scott Armstrong as presented at the Ninth International Conference on Decision Support Systems.  As such good forecasting requires a "tool-kit" of formulas to bring to bear, which necessarily needs to include the following major types of formula:

	Linear
	Non-linear
	Correlative

Business Intelligence tools are typically void of imbedded forecasting formula and, for those that do, it is usually limited to linear expressions.  However, most businesses are seasonal and the application of linear formula will more often produce an inaccurate and unusable forecast.

As mentioned, a forecast is a calculated value at a future point in time.  However, the future is full of uncertainties.   Now enter the Predictive element of Predictive Forecasting that applies probabilistic mathematics to assess the risk or confidence level in a forecast and provide a range of probable values.  To be clear forecasting and predicting are different though hand-and-glove related.  The weather report can be used to highlight this relation.  For example, the weather report of a "60% chance of rain tomorrow" contains a forecast "rain tomorrow" and a prediction "60% chance".  Forecasting and prediction when combined enable better decisions toward the effective deployment of capital and human resources and the mitigation of risks.

Presented below is an example of a Predictive Forecast that utilizes the Monte Carlo Simulation.  The white line represents the calculated forecast and the blue bars around the forecast line are the standard deviations or confidence intervals.  The darkest blue bar represents one standard deviation about the forecast or a 68% probability that the future outcome will fall into that range.  Each of the next lighter blue bar segments is two standard deviations with a 14% probability.  Each of the lightest blue bar segments is three standard deviations with a 1% probability the future outcome will fall into this range.  Rather than a point forecast the Predictive Forecast provides a quantitative assessment of a probable range of outcomes so that there is a 68% probability that sales will be between 590K to 578K units in the 3rd quarter.  Alternatively, the top of the darkest blue bar represents an 84% probability that the future outcome in the 3rd quarter will be below 590K units.

In benchmark after benchmark Predictive Forecasting has proven to be up to 98% more accurate, 90% faster and 70% less effort than Business Intelligence planning tools and spreadsheets.