Abstract: The traditional methods for construction land forecasting are based on static data which sim-ulate construction based on the trend extension method.However,when new data are acquired,thesemethods can not consider the relationship between new data and the original prediction results,andthus,construction land area needs to be reforecasted.This paper proposes a model which uses one-var-iable linear regression and least square method to correct prediction results dynamically when a type ofnew data is acquired.The results show that the corrected result is able to fit and replace the predictivevalues which are reprocessed using the new data.