I developed Data Forensics AI, a Python-based data analysis project focused on data quality analysis and anomaly detection.
I used statistical analysis and visualization to assess data quality, identify inconsistencies, and investigate anomalies.
- My analysis areas included missing-value assessment, duplicate detection, IQR-based outlier identification, categorical-data validation, correlation analysis, and regional sales and profit analysis.
I used Python, Pandas, NumPy, Matplotlib, and Seaborn for the project.
This project strengthened my practical understanding of data preprocessing, exploratory data analysis (EDA), data quality assessment, and insight generation to support data-driven decision-making.
I reflected that data is everywhere, but meaningful insights are not, and that data science's value is not simply processing information or building models.
I believe data science is valuable for its ability to question what we see, investigate what we overlook, and discover patterns that lead to better understanding.
Recently, I worked on developing a machine learning-based data investigation system that brings together anomaly detection, statistical analysis, clustering, and visual analytics.
The experience reinforced an important lesson: Finding patterns is technical. Understanding their significance is intellectual.
I see every project as an opportunity to move beyond simply using technology and start thinking about how it can solve meaningful problems.
While working on a machine-learning project in bioinformatics, I explored what we can discover when we visualise genetic variant data instead of looking at numbers in isolation.
I built a Python-based analytical workflow to explore this question through genomic feature analysis, anomaly detection, dimensionality reduction, and interactive visualisation.
- I used PCA, which projected multiple features into two dimensions to explore similarities and differences between variants; Forest, which flagged unusual feature combinations that might deserve further investigation; gene-level analysis to compare feature distributions across genes; and interactive visualizations to explore relationships between PrimateAI-3D, SpliceAI, PhyloP, and other features.
The current prototype uses synthetic data. An unusual pattern is not proof of pathogenicity, and a visualization is not a clinical conclusion.
The next challenge is to evaluate these ideas on curated, clinically labelled variants and determine which patterns are genuinely informative. For me, this project is about learning to ask better questions of biological data—not just producing charts or training another model.
Because in bioinformatics, finding a pattern is only the beginning. Understanding what that pattern means is the real work.
My approach is to keep continuously learning, experimenting, and building with purpose.
I describe urban environments as complex systems where mobility, infrastructure, and environmental indicators interact in different ways.
For a demonstration project, I explored urban datasets across six microzones using Python and data visualization.
- One focus was hourly mobility patterns across urban zones.
- I examined relationships between mobility and simulated air-quality indicators.
- I analyzed environmental variations through heatmap analysis.
- I compared zones using Comparative zone profiles using normalised radar charts.
The project reinforced my lesson that meaningful insights often emerge when multiple data dimensions are analyzed together.
The tools I used were Python, Pandas, NumPy, Matplotlib, Seaborn.
This exploratory project demonstrates my broader interest in using data visualization to understand how urban systems interact.