Skip to content
Github repo of a project used to fulfill my final graduation requirement for General Assembly's Data Science Immersive Bootcamp.
Jupyter Notebook
Branch: master
Clone or download
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
00_Datamunging, EDA
01_Bayesian Network
assets
.DS_Store
README.md
Shark Attacks.pdf

README.md

shark-attack-capstone

Github repo of a project used to fulfill my final graduation requirement for General Assembly's Data Science Immersive Bootcamp.

Executive Summary: Are you more or less likely to have an encounter with a shark if there were already two unprovoked encounters in that general location this year? My model aims to explain relationships between moon phase, occurance of last encounter, and type of encounter using a Bayesian Network.

Results:

  1. A model that can predict number of attacks per year given type
  2. A model that can predict type of attacks given number of attacks

What my model can tell you are conditional probabilities between the variables and I’m working on adding more. Further research will include activity, and if possible, adding the number of people enjoying the ocean in a given place at a given time.

Of the held out data we predicted 36 attacks total, for 90 held out observations, we correctly predicted the number of attacks there would be. 97% Accuracy.

If you remember my slide where we reviewed what influences shark attacks, only a few of those can very well be accounted for with data. (Policy, preventative measures, shark killing)

You can’t perform that action at this time.