Linear regression is a statistical method that can be used for forecasting trends. In contrast to most other forecast methods for trends the forecast parameters are not determined by starting with an initial assumption and then refining this assumption from one period to the next. Instead linear regression considers all the data at once and calculates a straight line through the data that results in the smallest error (sum of squares).
For more details on the statistical method, see Basics of Linear Regression.
Linear regression does not require any parameters, such as alpha or beta. The only parameter you can enter is a trend dampening profile.
An ex-post forecast is carried out so that you can compare the results with other methods.
No initialization is carried out. This means that you can use all the historical values to derive the forecast. As for all other forecast methods the more historical data you have the more accurate is the forecast.