Skip to content

juliachong/death-row-last-words

Repository files navigation

death-row-last-words

Goal: Categorize and predict death row inmates claiming innocence in their official last words statement by the race, age, race of victim, previous crimes, and education level through machine learning algorithms.

Data:

Process:

  • Used SMOTE to resample and create synthetic data points to combat the class imbalance, since innocence-claiming inmates were a huge minority.

  • Ran a decision tree after resampling, tuned and pruned the tree using a validation curve, gridsearch, and a randomforest.

  • Obtained the most important features using sklearn

  • Implemented K Nearest Neighbors and AdaBoost model

Conclusions:

  • K-Nearest Neighbors ended up being the most accurate model.
  • One of the biggest indicators of proclaimed innocence among all models was whether or not their victim was white.

About

☠️Guilt vs Innocence Clustering and Predictions

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published