Sign in Subscribe
Machine learning

Researchers Develop Machine Learning Model to Predict Precipitation in Iran

Strategic Insights

— 1 min read

Researchers at the University of Tabriz in Iran have developed a machine learning model using multilayer neural networks and decision trees to predict precipitation in the country's regions. The model, which was tested on a dataset of 24-hour cumulative precipitation data, showed a high level of accuracy in estimating monthly precipitation and classified Iran's provinces into seven clusters based on precipitation characteristics. The study's findings have significant implications for hydrological planning and water resource management in Iran.

Key Takeaways:

  • The machine learning model, which combines multilayer neural networks and decision trees, demonstrated high accuracy in estimating monthly precipitation in Iran, with a mean squared error (MSE) of 0.04.
  • The model classified Iran's provinces into seven clusters based on precipitation characteristics, with clusters 4 and 7 representing provinces with minimum and maximum precipitation, respectively.
  • The study found a significant effect of the time variable on precipitation variance, with a coefficient of 0.209.
  • The machine learning model was effective in analyzing hydrological data in Iran and can help improve precipitation forecasting systems.
  • The study's findings have significant implications for hydrological planning and water resource management in Iran.
  • The research was conducted by a team led by Mojtaba Fakhari, a PhD student at the University of Tabriz, with the assistance of Majid Rezaii Banafsheh Daragh, Behroz sarisarraf, Ali Mohammad Khorshid Dust, and Mozaffar Faraji.

Statistics:

  • 24-hour cumulative precipitation data was collected from Iranian stations and prepared using normalization.
  • The machine learning model, which included MLP with activation function s and decision tree with entropy criterion, was implemented to predict precipitation in Iran's regions.
  • The model's performance was evaluated and compared with accuracy, precision, and error criteria.
  • The mean squared error (MSE) of the model was 0.04, indicating a high level of accuracy in estimating monthly precipitation.
  • The coefficient of the time variable on precipitation variance was 0.209, indicating a significant effect of the time variable on precipitation variance.

Sources:

  • "Data Mining of 24-Hour Cumulative Precipitation Data in Iran Using Machine Learning: Multilayer Perceptron Neural Network and Decision Tree" by Mojtaba Fakhari et al. (akwhydrwlwzhy, 2025, 12(2): 795-811)
  • University of Tabriz (https://www.ut.ac.ir/)
  • Department of Climatology, University of Tabriz (https://www.utm.ac.ir/climatology/)
  • Information Technology Newsweekly (https://www.itnewsweekly.com/)
  • NewsRx LLC (https://www.newsrx.com/)

Read more