SharkSafe: Enhancing Diver Safety and Experiences Through Predictive Mapping of Shark Encounters

Overview The SharkSafe project empowers divers by providing detailed insights into shark encounters. Our goal is to enhance both safety and the quality of the diving experience through data-driven predictive mapping. We analyze shark encounter data to educate divers on optimal locations and times for diving, minimizing risks and maximizing enjoyment.

Project Scope Primary Hypothesis: Shark attacks are more likely to occur around midday, aligning with peak human activity, thus informing crucial safety advisories for divers. Secondary Hypothesis: Contrary to initial assumptions, visibility conditions in January are decreased due to extreme weather events, leading to higher risks despite increased shark encounters.

---

Data

This project utilizes shark encounter data that includes:

Date and Time Location and Geographical Data Details on Shark Species Environmental Factors and Human Activity

Difference Between DataClean and Seasonality Analysis

The project produces two main output datasets:

• shark_attacks_clean.csv (DataClean)
• seasonality_analysis.csv (Seasonality Analysis)

These files contain different levels of data detail.

---

1. DataClean Dataset (shark_attacks_clean.csv)

Description

This is the fully cleaned dataset, where each row represents a single shark attack.

It is the direct output of the cleaning pipeline:

Structure Example

Case_Number_final	Country	Month_name	Time_category	Fatal Y/N	Year_final

Characteristics

• Row-level data (granular)
• Contains all cleaned variables
• One row = one attack

Typical Uses

• Exploratory Data Analysis
• Statistical analysis
• Machine Learning
• Custom visualizations

Example

Country	Month_name	Time_category	Fatal Y/N
USA	July	Afternoon	Non-Fatal
USA	July	Afternoon	Fatal
Australia	January	Morning	Non-Fatal

Each row represents one individual shark attack.

---

2. Seasonality Analysis Dataset (seasonality_analysis.csv)

Each row represents a group of shark attacks.

Structure Example

Country	Month_name	Time_category	Fatal Y/N	Attack_Count

Characteristics

• Aggregated data
• Groups multiple attacks together
• Includes attack counts

Column: Attack Count

represents the number of attacks in each group.

Example

Country	Month_name	Time_category	Fatal Y/N	Attack_Count
USA	July	Afternoon	Non-Fatal	134
USA	July	Afternoon	Fatal	12

This means:

There were 134 non-fatal attacks in the USA in July during the afternoon.

---

Main Differences

Feature	DataClean	Seasonality Analysis
Data Type	Detailed	Aggregated
Level	Individual attack	Group of attacks
Rows	Many (e.g. 25,000+)	Fewer
Attack_Count column	No	Yes
Primary Use	General analysis	Seasonality analysis

---

Conceptual Difference

When to Use Each Dataset

Use DataClean When

• Detailed analysis is required
• Creating new visualizations
• Filtering records
• Exploring the dataset

---

Use Seasonality Analysis When

• Performing seasonal analysis
• Creating pivot tables
• Running fast aggregations
• Studying patterns by month or time of day

---

Summary

DataClean = individual attack records

Seasonality Analysis = aggregated attack counts

The cleaned dataset (shark_attacks_clean.csv) contains individual attack records, while the seasonality dataset (seasonality_analysis.csv) contains aggregated counts used for seasonal analysis.

---

Research Findings

Environmental Impact: Increased extreme weather events degrade visibility and heighten safety risks, Human Activity Correlation: Higher attack rates align with midday human activity peaks.

---

Contact

For questions or further collaboration, contact us at: info@sharksafe.com