Research Article
A Novel Fuzzy Time Series Forecasting Model Based on Multiple Linear Regression and Time Series Clustering
Algorithm 1
Learning algorithm of the proposed model.
Input: | | All multiple linear regression models ; | | Training sets ; | Output: | | Forecasting model with ANNs ; | (1) | ⟵ initialize all networks | (2) | ⟵ normalize | (3) | Repeat | (4) | For all do | (5) | For all do | (6) | Calculate the output value of | (7) | End for | (8) | , | (9) | ⟵ cost defined in equation (16) | (10) | For all do | (11) | Calculate the partial derivative defined in equation (19) | (12) | End for | (13) | | (14) | For all do | (15) | | (16) | Compute and for all weights and offsets | (17) | Update network weights and offsets | (18) | End for | (19) | End for | (20) | Until RMSE is small enough or fallen into a local minimum |
|