Analyses of the CANFIS, MLPNN, MLR, and M5 tree model for estimating the meteorological drought index (case study: Kermanshah synoptic station)

Document Type : Research Paper


1 Department of Science and Water Engineering, Imam Khomeini International University (IKIU)

2 Department of Water Engineering, Faculty of Agriculture and Natural Resources, Imam Khomeini International University, Qazvin, Iran.


This study examined the performance of the CANFIS fuzzy-neural model against other models, such as MLPNN, MLR, and M5 decision tree model, in predicting the SPI drought index in timescales of 12, 9, 6, 3, 1, and 24 months, for 70 months. The SPI index was developed by McKee and his colleagues at Colorado State University in 1993. A lack of rainfall can be calculated based on the probability of occurrence over a range of time periods from one month to 48 months utilizing this index. The optimal input was selected by using autocorrelation and partial autocorrelation analyses. In order to determine the relationship between the PACE charts and significant time delays for each index, MINITAB software was utilized to extract the PACE charts and calculate the significant time delays. In the context of different scenarios, the relationship between these variables is assessed using the CANFIS, MLR, MLPNN, and M5 tree models, ensuring that 70% of the data were used for training, 15% were checked for validation, and 15% were used for testing. CANFIS, MLR, MLPNN, and M5 tree models were evaluated by the root mean square error (MSE), root mean square error (RMSE), standard deviation (MAD), coefficient of determination (R2), and visual interpretation using scatter diagrams. In order to implement the CANFIS and MLPNN fuzzy neural models, the NeuroSolution software was used, and in order to model the M5 and MLR decision tree algorithms, the Weka software prepared by researchers at Waikato University was used.


Main Subjects

  • Receive Date: 02 November 2022
  • Revise Date: 24 June 2023
  • Accept Date: 26 June 2023
  • Publish Date: 26 June 2023