ezForecaster uses techniques known as Time Series Forecasting Methods to project future values.

Consider this example:

In January, we sell 2 products, and then in February, we sell 4 products. Time Series Forecasting looks for a pattern in the data and extrapolates it forwards. Given the evidence we have so far, we can guess two patterns:

Our sales are increasing by 2 every month - so in March we'll sell 6 products

or maybe

Our sales are doubling every month - so in March we will sell 8 products.

In this case, it's impossible to know which is the more accurate assessment - because in this deliberately simplified case, we don't have enough evidence, we need more historic data. Nonetheless, this example demonstrates the thought pattern within ezForecaster, which is to try every possible method and see which one or ones are most appropriate and project them forward.

In fact, ezForecaster has 21 different forecasting techniques, and it runs some of them hundreds of times over with different settings, so in a second, ezForecaster will literally have tried out thousands of pattern matching methods and will have determined the best one.