The term forecasting software is quite ambiguous because it can refer to very different things. This post is a modest attempt to clarify the situation.
Deterministic simulation softwareA common type of ''forecasting'' software is the simulation software. As input, the software takes a set of rules, of facts, of constraints and this information is used to run a deterministic simulation. The results of the simulation constitute the ''forecasts'. The simulation is said to be deterministic if the same input data produces exactly the output results. In such simulation, there is no ''uncertainty'' and no ''randomness''.
Many staff scheduling software are simulation forecasting software: a schedule must be made for each employee, each schedule must respect many constraints and the resulting schedule combination must also respect constraints).
Expert insights aggregation softwareContrary to the simulation situation where all parameters are ''under control'', there are many business situations that depend on events that occur with a proportion of chance / randomness. A typical way of handling the business uncertainty consists in relying on people's expertise to figure out what will happen next. Many software are designed to aggregate the insights delivered by the ''experts'' of the company to deliver a global forecast.
Many CRM (Customer Relationship Management) software provide expert aggregation forecasting. Each vendor (the experts) provide a probability of success for his own upcoming deals. The CRM software computes the global upcoming sales figures by summing all the vendor's forecasts.
Statistical forecasting softwareBeside the company experts, another very valuable source of information to predict the future is the historical data of the company. Statistical methods can be used to produce forecasts automatically based on the historical data. Statistical forecasting software rely on those mathematical methods to produce their forecasts.
Most supply chain optimization software relies on statistical methods to forecast the inputs/outputs of products. Statistical methods can also be used to anticipate customer demand (ex: future call volumes for a call center) and adjust staffing capacity accordingly.
The Lokad Case
Lokad is a statistical forecasting software. As discussed here above, Lokad relies on historical data and not on any expert insights. In addition, Lokad is mostly an hosted application. The forecasts are computed directly on the Lokad.com servers and not on the customer local machine. This approach leads to a greater forecasting accuracy.