The FastForward product is a time series forecast engine that uses a highly customized Neural Network to produce excellent forecasting results. It is highly optimized and efficient. It was designed to integrate easily into new or existing applications. It is self-tuning and easy to maintain since it is self-contained and doesn’t require any third party or open source libraries. It’s architecture makes it highly portable and scalable.

Artificial Neural Network (ANN) technology is a form of artificial intelligence that is inspired by the functioning of the human brain. By creating an interconnected group of artificial neurons an adaptive system can be created that provides an excellent non-linear statistical data modeling tool that can be used for regression analysis (data mining of patterns). Similar technologies are used in biometric systems such as finger print and face recognition. These types of systems have improved greatly over the last decade.

The FastForward forecast engine is a non-recurrent multilayer feed forward neural network that uses nonlinear weighted sums. It is highly optimized and efficient to prove fast and accurate forecasts.

Many of the competitors to the FastForward product use older technologies that are less efficient and don’t scale well due to the inclusion of older libraries (i.e. Fortran). Linear forecasting methods such as ARIMA (Autoregressive Integrated Moving Average) are more sensitive to missing data points then a non-linear approach. These missing input data points can cause problems in these models. Linear approaches do not forecast well for non-linear data.

The FastForward forecast engine was designed to provide excellent forecasts right “Out of the box”. Very little external configuration or tuning is required to use the product. This allows developers to easily integrate forecasting capability into their applications without having to become experts in regression analysis. FastForward does allow tuning of many internal parameters but in most instances no manual tuning is necessary.

One of the unique outputs of the FastForward forecast engine is the forecast’s certainty factor. This value represents a degree of confirmation and is not a probability as in a Bayesian sense. The certainty factor is a value that represents the forecast’s relative accuracy as compared to other forecasts made by the forecast engine. It varies with each forecast based on the input data and the forecast horizon.

The certainty factor allows the consumer of the forecast to make decisions about which forecasts to trust automatically and which should be inspected or adjusted by experts. For example, an inventory forecasting application could use the certainty factor and the item profitability value to sort the forecasted items in the order that they should be reviewed by purchasing agents. This would allow the agents to maximize their effectiveness.

FastForward is very fast.

The FastForward engine was designed to be easily integrated into your application. It is written in Java which makes it highly portable. Java is supported on many platforms from cell phones to mainframes. The forecasting engine is typically integrated directly with your application as a Java jar file. The ability to roll it into your Java application means there isn’t a separate installation procedure or remote communication issues or security concerns. It runs in the same process and is simply another library that your application calls.

Easy integration also means that your development team can get the application up and running sooner with fewer issues. This translates into a faster deployment time and fewer code defects which reduces development costs.

FastForward is self-tuning and adaptive. It learns over time and can modify its behavior on a per forecast basis. This ability is crucial because it is often impractical to hand tune and constantly manually adjust forecasting parameters for every forecast entity. The Self-Tuning ability of FastForward sets it apart from the competition.

Using FastForward simplifies application maintenance. Since the forecast engine is contained in a single jar file it is easy to upgrade. This results in fewer upgrade issues and a better customer experience and reduced operational costs.

FastForward was designed to deliver excellent forecasts “Out of the Box”. It is self-tuning. This means you don’t need to be an expert in mathematical modeling to use it effectively. As the forecasts improve you are able to run your business better and improve your profitability.

The FastForward forecast engine can except and detect correlations in multiple input series. These additional independent series can be used to improve the forecast accuracy. For example, an ice cream store could provide the forecast engine with the daily local temperature for the store location as well as the ice cream sales data for that store. The forecast engine will determine the strength of the correlation between the temperature and the ice cream sales. By providing the future expected temperatures for the store location the accuracy of the forecast for those stores are improved.

Another application for adding independent series would be running sales promotions and gauging their impact on sales. Different types of promotional data can be feed as input to the forecast engine which can analyze the effect of the promotion on sales. The forecast engine can then do forecasts with and without the promotions in the future. This is useful for “What if” scenarios.

Java technology provides mechanisms for writing reliable code. It’s object oriented nature, lack of pointers and its use of built in design patterns make it well suited for enterprise level software. The FastForward product does not depend on third party libraries or communication protocols with other servers. Other solutions have various other libraries and components that must be installed and maintained in order for the forecast engine to function properly. Often these individual components have their own versions and maintenance procedures.

Since the forecast engine is written in Java it is highly portable and the same jar files can run on many platforms. This flexibility can be important to companies that run multiple operating systems on dissimilar hardware.

The FastForward forecast engine is fast. Chances are it’ll run very well on hardware your company already owns or it’ll require less computing horse power then other similar options. When forecasts computational time is an issue then FastForward wins hands down.