| 要約: | This study describes application of a hybrid combination of hybrid evolutionary algorithm (HEA) and thematic map visualization technique in modeling, predicting and visualization of selected algae division growth, Chlorophyta for tropical Putrajaya Lakes and Wetlands (Malaysia). The system was trained and tested using five years of limnological time-series data sampled from tropical Putrajaya Lake and Wetlands (Malaysia). HEA is run on the training set in order to provide insights on the relationships between input variables and the algae abundance. Performances of the rule sets are assessed using Receiving Operating Characteristic (ROC) with true positive rate. The generated rules are tested with another set data to avoid biasness, which yielded accuracy rate of 73%. The rules generated by HEA are then integrated with thematic map technique for visualization of the Chlorophyta abundance. Input parameters are optimized using HEA to weed out insignificant input for predicting Chlorophyta abundance. The optimized variables are namely rainfall, wind speed, sunshine, temperature, pH, dissolved oxygen, Secchi, turbidity, conductivity, total phosphorus, ammonia (NH3-N), nitrate (NO3-N), biochemical oxygen demand, chemical oxygen demand and total suspended solids.
|
|---|
