/Predicting-Mosquito-Hotspots

State of the art prediction of mosquito hotspots and virus presence using machine learning and deep learning

Primary LanguageJupyter Notebook

Predicting the next mosquito hotspots using past data.

As an assignment from General Assembly Data Science Immersive School

Objective

Using data. Find patterns and predict the next mosquito hotspot areas and whether disease would be present.

Machine Learning Frameworks used

  • SKLearn

Machine Learning Applications:

  • Feature Engineering
  • Feature Selection using random forests
  • Validation / Cross-Validations
  • Hyperparameter tunings, GridSearch
  • Model Selection

Ensemble and model stacking

I apply cross validation and perform model ensemble for various models then use a meta learner to learn the output of these models to achieve an even higher predictive accuracy.

  • Used an ensemble of random forest classifiers and sklearn without having to resort to deep learning and yet still achieved a 99% accuracy.