README_en.md

Model

Train and Validate Data

Train and Validate data ("train data" for short) is divided into 3 group according to rain amount to train classification model (RF1):

• Group 1: No rain (0 mm/h)

• Group 2: Weak rain (under 2.8 mm/h)

• Group 3: Strong/Heavy Rain (over 2.8 mm/h)

After that, these data is copied into 2. The first one is for training classification model (RF1). The second one will be divided into 3 part corresponding to 3 group of rain. Part 2 and 3 (Group 2 and 3) will be used for training RF2 and RF3

Because of the data is imbalanced, train data will be re-balanced by SMOTETomek-links technique

Tunning model

These hyperparemeter will be adjust for tunning process

• n_estimators: range(100, 3000, 100)

• max_features: range(0.05, 1.0, 0.05)

• min_samples_split: range(0.025, 0.5, 0.025)

• min_samples_leaf: range(0.05, 1.0, 0.05)

• max_samples: range(0.05, 1.0, 0.05)

• min_weight_fraction_leaf: range(0.025, 0.5, 0.025)

Run

Run RandomForest.py to Estimate Rainfall

Modify dataset location in ImportData.py

Code in Jupyter Notebook files are similar to code in python file, except ImportData.py

Jupyter Notebook folder missing ImportData.py