This repository contains two applied data analysis projects focused on exploratory visualization and network analysis using real-world datasets. Both projects emphasize the use of Python libraries for data cleaning, statistical analysis, and graphical representation of insights for both technical and non-technical audiences.
The goal of this project was to analyze animal intake and outcome data from the Austin Animal Shelter (Texas, USA) to understand adoption patterns and seasonal trends. Through exploratory data analysis and visualization, the project sought to identify key relationships between animal type, age, and adoption outcomes.
-
Data processing: Cleaned and merged two datasets (
intakes.csvandoutcomes.csv), handled duplicates, missing values, and inconsistent entries. -
Feature engineering: Converted date and age fields to numerical and datetime types for statistical exploration.
-
Visualization: Used matplotlib, seaborn, and pandas plotting for univariate and bivariate analysis.
- Adoption trends by month and animal type.
- Correlation between intake type and adoption success.
- Age distribution of adopted vs. non-adopted animals.
- Identified seasonal peaks in animal intake and higher adoption rates for younger animals.
- Discovered a strong relationship between intake reason and outcome type.
- Presented findings through clear, annotated visualizations suitable for public data storytelling.
This project aimed to perform network analysis of Prague’s Integrated Transport System (PID) using stop-to-stop transit data. The main goal was to visualize and analyze the connectivity and centrality of public transport stops within the city.
-
Data preprocessing: Cleaned and structured CSV data derived from GTFS (General Transit Feed Specification), adjusted time formats, and unified stop identifiers.
-
Network construction: Modeled stops as nodes and connections as directed edges weighted by weekly service frequency.
-
Analysis:
- Computed degree, closeness, and betweenness centrality using NetworkX.
- Visualized subnetworks for trams and buses separately to highlight key transport hubs.
- Identified most central stops in Prague’s transport network based on traffic volume and connectivity.
- Visualized complex transit relationships in a readable, interactive graph structure.
- Gained insights into network resilience and load distribution across transport modes.
- Languages: Python 3.x
- Libraries: pandas, numpy, matplotlib, seaborn, NetworkX
- Tools: Jupyter Notebook, Git, GitHub