jade-bejide/datathon-2023

Our Submisson for the BDSS x LV= 2023 Datathon

datathon-2023

Table of Contents

• datathon-2023
  • About the Datathon
  • Team
  • Objectives
  • Dataset
  • Preprocessing
  • Modelling

---

About the Datathon

This Datathon was hosted by the Bristol Data Science Society (BDSS) in association with LV=. The format of the competition was a data science oriented Hackathon where we were given a real world dataset and a prediction task.

---

Team

Our team (called work in progress) consists of 2 people from the University of Bristol:

• Nikhil Parimi (Computer Science, 2nd Year)
• Jadesola Bejide (Computer Science, 2nd Year)

---

Objectives

Predict whether a fatal/serious casualty occurs

Predict the casualty_severity column in the casualty_test.csv dataset for whether an accident is fatal/serious or slight using road traffic data about the casualty, accident, and vehicles involved.

Note: Map targets to binary 0 (fatal, serious) and 1 (slight).

---

Dataset

1. Casualty (primary)

• Data on the person(s) involved in the casualty; casualty_train.csv

2. Vehicle (secondary)

• Data on the vehicles involved in the casualty; vehicle_train.csv

3. Data Dictionary (reference)

• Field descriptions and value mappings; DatathonReference.xlsx

---

Preprocessing

1. Merged both csv files together on common field accident_reference to make detailed statistical analyses about the features
2. Merged the values of the column casualty_severity into binary form; mapping 1 to fatal / serious and 0 to slight.

Modelling

1. Dropped features with no gaussian or monotonically increasing/decreasing correlation
2. Used scikit-learn to train various machine learning models, including Naive Bayes and Support Vector Machines