PT-BR

Se você quiser ler em português basta clicar aqui

📊 Catboost Model Dashboard

🎯 Main Objective of the Repository

Simplify the visualization of Machine Learning (ML) models using an interactive dashboard.

🚀 Features

• 📁 Model Upload: Allows the user to upload a JSON file with the model parameters.
• 📈 Data Visualization: Uses data from data/assets/asset_data.parquet to create ML models with the CatBoost algorithm.
• 📊 Generation of Charts and Tables: Produces result charts, metrics (Drawdown, Sequential Results, Win Rate, and Expected Return), and recommendation and sequential result tables.

🛠️ Usage Instructions

Step 1: 📄 Create the JSON File

Create a JSON file containing all the necessary parameters for the base_model_creation function located in machine_learning/model_builder.py.

Step 2: ⬆️ Upload the Model

In the dashboard, click on "Upload Model" and select the JSON file created in the previous step. The dashboard will use the data from data/assets/asset_data.parquet to create the ML model using the CatBoost algorithm.

Step 3: 📊 View the Results

After the upload, the dashboard will automatically generate:

• Result Chart
• Metrics Chart: Includes Drawdown, Sequential Results, Win Rate, and Expected Return.
• Recommendation and Sequential Result Tables

📂 Repository Structure

• data/: Contains the data used to train the models.
• machine_learning/: Contains scripts for model creation and mining.
  • model_builder.py: Main script for model creation.
  • model_miner.py: Facilitates the search for the ideal ML model configuration.
  • model_features.py: Simplifies the creation of features for the CatBoost algorithm.
• run_model.py: Main script for model execution and visualization generation.

📋 Not Implemented (To-Do for Future Projects)

Model Miner

Although not implemented in the dashboard, the machine_learning folder has a file called model_miner.py that facilitates the search for the ideal ML model configuration. The final result will be a dict that, when transformed into JSON, will be compatible with the site's visualization. It is possible to perform model mining using a Jupyter Notebook file and use the dashboard to track results.

Model Features

Another relevant file is model_features.py, which has a class that simplifies the creation of features to be used in the CatBoost algorithm.

📦 Dependency Installation

To download all the dashboard dependencies, use the command:

pip install -r requirements.txt