This device is not compatible.


Predict Frog Toxicity with Python and XGBoost

In this project, we’ll learn to use the gradient boosting algorithm in Python by building a model to predict the toxicity of poison dart frogs by their color and luminance.

Predict Frog Toxicity with Python and XGBoost

You will learn to:

Implement a gradient boosting model.

Clean and label research data.

Tune and improve a baseline model.

Export a fully trained model.


Machine Learning

Data Cleaning

Data Science

Data Visualization


Familiarity with Python

Basic understanding of working with data in Python

Basic understanding of machine learning principles




XGBoost logo



Project Description

Poisonous or toxic organisms often present bold colorations and flashy patterns, which serve as a defense mechanism. The bold colors serve as an advertisement for their toxicity, shouting out “Don’t eat me!” to any potential predators. Although there’s not a lot of concrete research on the correlation between visual warning signals and animals’ toxicity, a research paper was published in The American Naturalist in 2011 that detailed how the color and brightness of a certain species of poison frogs are reliable indicators of their toxicity levels.

image image image

In this project, we’ll learn how to build gradient boosting machine learning models in Python by building a model that can predict a frog’s toxicity levels by its luminance.

We’ll start with using pandas to load and clean the data. Then, we’ll visualize it using seaborn and Matplotlib to gain a better understanding of the data and how we can use it to train the model. Next, we’ll train a baseline XGBoost model, perform simple optimizations, and analyze the performance. Afterward, we’ll conclude the project with the final step of saving the model using ONNX.

Project Tasks


Data Processing

Task 1: Import Libraries

Task 2: Load the Dataset

Task 3: Clean the Data

Task 4: Perform Simple EDA

Task 5: Split the Data


Model Training

Task 6: Train a Baseline Model

Task 7: Tune the Model

Task 8: Compare Model Performance


Model Export

Task 9: Prepare the Model for Export

Task 10: Export the Model