Introduction

Hyper-parameters optimization is an important step to build Machine Learning solutions. Random Forest Grid Search simulator helps the user to understand how the results of a classification model are being affected after changing the algorithm hyper-parameters.

Shiny app backend characteristics:

• Machine Learning Model: Random Forest.

• Dataset: Subset of 20.000 random observations from Australia Rain dataset.

• Train/test split: 80% training - 20% test.

• Chosen predictors: MaxTemp and Humidity3pm.

• Hyper-parameters to tune up: number of trees and maximum tree depth.

Code for Shiny app and this pitch can be found on Github.

User Interface

User instructions

The simulator is easy to use! Just follow the steps below:

1. Select number of trees.

2. Select maximum tree depth.

3. Click on “Train model”.

4. Enjoy!

Result | Decision boundary plot