Jonathandav/Bird_Migration_Analysis

---
language:
- en
license: mit
size_categories:
- 10k<n<100k
task_categories:
- tabular-classification
pretty_name: Bird Migration Survival Analysis
tags:
- ecology
- biology
- wildlife-tracking
- data-science
---

# Bird Migration: Identifying Survival Corridors

## Project Overview
This project investigates a dataset of 10,000 migratory bird records to identify the physiological and environmental factors that determine migration success. By leveraging exploratory data analysis (EDA) and advanced statistical visualization, this study identifies a specific "Safe Zone" for flight altitude and challenges common assumptions regarding resting behaviors.

---

## Principal Insights & Visualizations

### 1. The Key Factors
I started with this wide-angle view of all 42 variables (see chart below), which then led me to focus on Altitude and Rest Stops for my deeper dive.


### 2. The Altitude "Safe Zone"
Through Kernel Density Estimation (KDE) analysis, i was able to identify a clear survival corridor. Birds maintaining a **minimum altitude between 600m and 900m** show a significantly higher density of successful migrations compared to those flying at lower altitudes.


### 3. The Rest Stops (Efficiency vs. Effort)
An analysis of behavioral patterns reveals that the number of rest stops taken (1–14) does not statistically correlate with success. this suggests that migration efficiency is driven by in flight strategy rather than the quantity of grounded rest.


### 4. Geographic starting point
Using Violin Plots to analyze starting latitudes, the data shows nearly identical distributions for both successful and failed migrations. This further indicates that survival is determined mainly by choices made during transit (such as altitude) rather than the geographic point of origin.


---

## Data Dictionary
| Category | Features | Description |
| :--- | :--- | :--- |
| **Target** | `Migration_Success` | Binary outcome (1 = Success, 0 = Failed) |
| **Flight Dynamics** | `Min_Altitude_m`, `Average_Speed_kmph` | Key physical metrics recorded during transit |
| **Environment** | `Temperature_C`, `Humidity_%` | Atmospheric conditions at the time of migration |
| **Behavioral** | `Rest_Stops`, `Predator_Sightings` | Actions taken and risks encountered by the bird |

---

## Data Wrangling & Methodology
To ensure a high-quality analysis, several technical challenges were addressed:
* **Label Standardizing:** Resolved a significant data quality issue where migration outcomes were inconsistently labeled (Successful/Failed vs Yes/No) and converted into .
* **Outlier Strategy:** Investigated 86 high-speed outliers (>180 km/h). These were retained as they represent valid biological "burst" flight events rather than sensor errors.
* **Class Balancing:** Verified a balanced distribution of ~5,000 successful and ~5,000 failed migrations to prevent model/analysis bias.

---

## Future Work
* **Hardware Analysis:** Investigating the correlation between `Tag_Battery_Level_%` and reported failure to distinguish between biological failure and equipment failure.
* **Conservation Modeling:** Utilizing the identified 750m optimal altitude to map protected air corridors for endangered species.

---

* **Important note:** The dataset i used was a sythentic dataset, which made the stats inconsistent with reality. I decided to analyze the data as if it were real, so results do not reflect reality.

---

**Developed by Jonathan David** *Tools used: Python, Pandas, Matplotlib, Seaborn, Google Colab.*