Exoplanet Prediction in Deep Space using DeepNet and Advanced Ensemble

Objective

Predict whether a planet is an exoplanet or not.

Definition

An exoplanet is any planet in other solar systems, and its presence is determined by analyzing the flux of the star it is revolving around.

Algorithms Used

• Deep Sequential Model (ANN): Utilized Keras Tuner for hyperparameter tuning, achieving an accuracy of 51.62%.

• Advanced Ensemble:

  • XGBoost: Hyperparameter tuned to achieve an accuracy of 96%.
  • LightGBM: Hyperparameter tuned to achieve an accuracy of 97.19%.
  • HistGBM: Hyperparameter tuned to achieve an accuracy of 97.45%.

Tuned Parameters

Tuned parameters are stored in the 'Tuner Params' directory (zipped) and can be used to study the architecture of the DeepNet model.