https://github.com/sckonung/crab-age-regression

ML model for regression with a crab age dataset Competition in Kaggle
https://github.com/sckonung/crab-age-regression

keras machine-learning pandas python scikit-learn tensorflow

Last synced: about 1 month ago
JSON representation

ML model for regression with a crab age dataset Competition in Kaggle

• Host: GitHub
• URL: https://github.com/sckonung/crab-age-regression
• Owner: scKonung
• License: mit
• Created: 2024-08-23T12:52:24.000Z (5 months ago)
• Default Branch: main
• Last Pushed: 2024-08-23T13:19:58.000Z (5 months ago)
• Last Synced: 2024-11-03T03:43:01.412Z (3 months ago)
• Topics: keras, machine-learning, pandas, python, scikit-learn, tensorflow
• Language: Jupyter Notebook
• Homepage: https://www.kaggle.com/competitions/playground-series-s3e16/overview
• Size: 8.79 KB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Crab Age Prediction Model

This project marks my debut in Kaggle competitions, where I developed a straightforward model to predict the age of crabs.

## Dataset Overview

The dataset I utilized for this model is sourced from [this dataset](https://www.kaggle.com/datasets/sidhus/crab-age-prediction). To enhance the model's input, I applied One Hot Encoding to the 'Sex' column, creating individual columns for each sex category.


**Original Data Snapshot:**


![Original Data](https://github.com/user-attachments/assets/b5f722de-80a5-4a35-b1ac-31a017afa7f8)


**Encoded Data Snapshot:**


![Encoded Data](https://github.com/user-attachments/assets/3bbf0b97-d16f-4999-add3-3eb5dc1d91ef)


## Model Overview

The model architecture I employed consists of three Dense layers, with the first two layers having 32 output units each, and the final layer outputting a single value for age prediction. I utilized ReLU activation functions throughout.

## Model Training and Optimization

For training optimization, I chose the RMSprop optimizer and Mean Squared Error (MSE) as the loss function. During the competition phase, I trained the model over 100 epochs with a batch size of 32.

This approach allowed me to create an effective baseline model for predicting crab ages, leveraging straightforward yet powerful techniques to tackle the problem at hand.