Othon Viannis

About

As an aspiring data analyst, I bridge the gap between raw information and human understanding. What started with structuring data pipelines in Excel and Power Query has grown into a practice of crafting interactive visualizations using Python, QGIS, and Flourish—each project an opportunity to reveal the narratives hidden within datasets.I'm drawn to the messy, the complex, the overlooked. Whether mapping global migration patterns or uncovering cultural heritage through numbers, I believe every dataset holds a human story waiting to be told. My approach combines analytical precision with thoughtful design, creating dashboards and visuals that don't just present information—they invite exploration and spark insight.Currently, I'm seeking opportunities where I can apply my skills in data management and visualization to create meaningful impact. Let's connect—explore my projects below or reach out via [Contact/LinkedIn].

• Back

Developed a comprehensive health analytics simulation combining VBA automation, descriptive statistics, and correlation modeling to explore the behavioral and clinical factors influencing diabetes risk

In this project I simulated an end-to-end public health analytics workflow, integrating data entry, statistical analysis, and visualization using a dataset of over 250,000 anonymized health records.It features automated VBA forms for streamlined patient data input and personalized risk calculations, alongside PivotTable dashboards that visualize behavioral and lifestyle risk patterns.Correlation and regression analyses quantify the impact of BMI, mental health, and physical health on diabetes outcomes.The system culminates in automated PDF report generation, simulating professional healthcare reporting.

The goal was to simulate a complete health analytics workflow — from raw patient data to diagnostic dashboards and statistical modeling — using Excel as a single, unified analysis environment.

Business Questions Answered:

Data Structure:

The dataset contained 16 key health and lifestyle indicators, including:Diabetes, Diabetes Risk Score, Cholesterol Levels, BMI, Smoking (Yes/No), Stroke (Yes/No), Heart Disease, Physical Activity (Yes/No), Fruits (Yes/No), Veggies (Yes/No), Alcohol Consumption (Heavy/Not Heavy), General Health, Mental Health, Physical Health, Difficulty Walking, Sex, and Age.

Analytical Components

PivotTables & Dashboards

Five analytical dashboards were created, each powered by dedicated PivotTables:

Each dashboard was visualized using bar and donut charts, supported by conditional formatting for risk highlighting.

Exploratory Data Analysis

Decriptive Statistics

I studied BMI, mental health, and physical health in a dataset of 253,679 people:

Regression Statistics

The model shows a moderate connection between these factors and diabetes risk, explaining a small but meaningful portion of why risk varies. With so many records, the results are reliable.The model confirms that BMI, mental health, and physical health play a role in diabetes risk, though other factors also matter.

Each Factor:

What It Means: Most people are overweight, and some have notable mental or physical health issues, which could increase diabetes risk, especially for those with higher BMIs.
Being overweight is the biggest factor driving diabetes risk. Poor mental and physical health add smaller risks. This supports focusing on weight management and overall health to lower diabetes chances.

Correlation Matrix

A correlation matrix was developed to explore the relationships among BMI, Mental Health, Physical Health, and Diabetes Risk.

Key Insights:BMI shows an almost perfect correlation with diabetes risk, confirming it as the primary clinical predictor.
Mental and physical health factors exert smaller yet significant effects — highlighting that while diabetes is largely metabolic, behavioral and psychological dimensions contribute modestly to overall risk.

Comorbidity Analysis

Assessed the overlap of diabetes with major conditions:

Insight: Diabetes, hypertension, and obesity often co-occur — forming a chronic disease triad that requires integrated intervention.

Behavioral Risk Heatmap

Assessed the overlap of diabetes with major conditions:

Insight: Smoking remains the stronger predictor of diabetes.
The lower prevalence among heavy drinkers likely reflects demographic skew or data artifacts, not causation.

Key Insights

Conclusion

The analysis highlights BMI as the dominant predictor of diabetes risk (r = 0.974), showing a strong, near-linear link where higher body mass significantly raises the likelihood of developing diabetes. This underscores weight management as the most essential preventive strategy.
In addition, physical and mental health play smaller but significant roles, with more days of poor physical health and even slight mental health declines linked to elevated risk. These findings reveal diabetes as a multifaceted condition influenced not just by metabolic factors like BMI, but also by lifestyle behaviors, activity levels, stress, and emotional well-being.

• Back

*Developed a comprehensive analytics simulation using a manually curated dataset, gathered from official UK retailer and brand websites

This project utilizes a manually curated dataset collected from 14 official UK retailer and brand store websites for 35 different smartphone models across six major brands: Google, Honor, Motorola, Samsung, Xiaomi, and OnePlus.I manually gathered all data (prices, ratings, reviews, and availability) directly from each site and thoroughly cleaned the data in Excel by standardizing formats, correcting model names, and removing duplicates to ensure integrity. I then created specialized calculated fields, including a Value Ratio (Rating / Price) and an Engagement Ratio (Reviews / Price), to enable advanced comparative analysis.

The goal was to simulate a complete data workflow — from manual data collection to a dynamic Excel dashboard — demonstrating how real-world market insights can be derived.

Business Questions Answered:

Data Preparation:

For this project, I manually gathered data from 14 official UK retailers and brand stores. The data was thoroughly cleaned in Excel by standardizing price formats, correcting brand and model names, and removing duplicates to ensure integrity.I then created specialized calculated fields, including a Value Ratio (Rating / Price) and an Engagement Ratio (Reviews / Price), and verified all source links and retailer entries for consistency and completeness.

Pivot Tables

I generated ten core PivotTables to structure the comparative analysis of the smartphone data.

The PivotTables established the foundation of the analysis by aggregating and segmenting the data, allowing for direct comparison of metrics.

Dashboards & Visualization

The analysis was presented through four dashboards created for clear comparative analysis. Key visualizations included:

Key Insights

Conclusion

This project demonstrates how manual research and Excel-based analysis can generate meaningful insights into competitive markets.
By combining self-collected data with structured PivotTables and dashboards, the analysis highlights real-world consumer and pricing trends

• Back

Made for an internal e-commerce analytics simulation. Recreated with anonymized Amazon Fashion sales metrics.

Developed a structured Excel workflow and dashboard to consolidate and analyze over 128,000 rows of monthly Amazon Fashion sales across multiple product categories and seasons. Replaced manual reporting with automated formulas and PivotTables, enabling refreshable insights into revenue, returns, and customer trends over a 12-month period.

The goal was to simulate an end-to-end workflow — from raw e-commerce data to business-ready visual insights.

Business Questions Answered:

Data Processing Steps:

Firtly, I imported a 2022 CSV Amazon Fashion dataset from Kaggle containing 128,974 rows of transaction records.Then, I cleaned and standardized data fields by removing duplicates, handling null values, and standardizing columns for date, category, and currency.

This analytics project involved comprehensive data visualization and summarization using PivotTables and PivotCharts.

The pivot tables cover:

The analysis provided deep-dive metrics on revenue/return rates per category and average customer value, featuring slicers for filtering and conditional formatting to highlight performance peaks.

Additionally, I built three PivotCharts to visualize trends and performance:

Key Insights

Conclusion

This project demonstrates practical Excel data analysis skills applied to large-scale e-commerce data.
By transforming raw sales data into a visually clear dashboard, it simplifies performance monitoring and supports better business decisions.

• Back

Developed as a web data automation project demonstrating Excel’s advanced Power Query and formula-based transformation capabilities.

In this project I built a fully automated Excel workflow that extracts live job listings from RemoteOK, cleans messy web data, and structures it into an analysis-ready table. The project demonstrates an end-to-end workflow — from live web import to column extraction and cleaning — .

The goal was to show how Excel can act as a no-code web scraping and data transformation tool for real-world applications, such as job market research and remote work listings.

Business Questions Answered:

Data Processing Steps:

Data Extraction (Power Query – From Web)

Data Transformation (Power Query)

Data Enhancement (Excel Formulas)

Techniques Used:

Power Query (ETL Workflow):

Excel Functions:

Key Insights

Conclusion

This project demonstrates how Excel can serve as a lightweight data scraping and cleaning platform without any external scripts or APIs.By integrating Power Query for data import and Excel formulas for text parsing and enhancement, the workflow converts raw, unstructured web content into a clean, ready-to-analyze dataset.The result is a practical example of no-code web data automation, bridging everyday spreadsheet use with data analytics functionality.

• Back

This project explores the evolution of Swedish playwright August Strindberg's works from naturalism to expressionism, using digital humanities tools to uncover psychological and sociocultural patterns in 29 plays.By analyzing a corpus of naturalistic (realistic, social-focused) and expressionist (symbolic, emotional) plays, I examined how Strindberg's personal crises influenced his themes, blending psychoanalytic (inner conflicts) and constructivist (societal norms) perspectives.

Key Mehods Included:

Topic Modeling: Using Python's Gensim library and LDA algorithm, I identified 5 topics per style. Naturalistic plays emphasized family dynamics and social issues (e.g., "marriage," "master"), reflecting 19th-century realism. Expressionist plays focused on existential motifs (e.g., "prince," "witch"), symbolizing inner turmoil and alienation.

To clearly understand the discovered topics, the pyLDAvis library was used to create an interactive visualization that intuitively displays topic distributions and word relationships. This was supplemented with matplotlib bar plots to precisely show the top words and their probabilities for each topic.

Named Entity Recognition (NER) and Relationships: Analyzed entities (locations, organizations, persons) with normalized frequencies. Expressionist plays showed broader diversity (e.g., more GPE like "Jerusalem"), indicating symbolic exploration.

Selective list of Naturalistic plays entities:

Selective list of Expresionist plays entities:

A co-occurrence analysis was performed to uncover relationships and associations between entities. This data was used to construct network graphs via matplotlib, where entities (like people, locations, or organizations) are nodes, and their frequent joint appearances in the text are weighted edges. The resulting graph highlights key entities and their importance within the narrative by analyzing the centrality of nodes and the density of their connections

Sentiment Analysis: Applied VADER and TextBlob to gauge emotional tones. Naturalistic plays had a balanced sentiment (score 0.0730), while expressionist ones were more variable (0.0592). Timeline mapping to Strindberg's life events (e.g., Inferno crisis) linked low scores to personal struggles, like in "The Stronger" (post-divorce).

To more comprehensively explore the evolution of emotional dimensions in Strindberg's plays, the sentiment analysis was complemented by a temporal analysis of sentiment scores. This temporal method contextualized and visualized the results by correlating sentiment scores with the years the plays were written, effectively revealing patterns of emotional expression across his entire career.

• Explore Code on GitHub

Accompanying the analysis I created a video timeline visualizing Strindberg’s life and creative journey.
This video visualizes August Strindberg’s life (1849–1912) and creative evolution, blending spatial mapping with cinematic storytelling. Using QGIS, I mapped Strindberg’s residences and publication cities, with green dots marking his locations over time. DaVinci Resolve was used to craft a dynamic timeline, where flashes highlight major play publications and key life events.
This fusion of geographic data and biography transforms archival records into a vivid narrative, showcasing Strindberg’s journey from naturalism to expressionism.

This project merges computational text analysis with visual storytelling, demonstrating how data-driven methods can illuminate the emotional and intellectual evolution of a writer. By combining literary study with digital mapping and video, it bridges humanities research with creative digital visualization.

• Back

The Syrian civil war, erupting in 2011, displaced over 13 million people, sparking the 21st century’s largest refugee crisis.This project weaves a vivid tapestry of their journeys, using QGIS and Flourish to map and visualize the crisis’s patterns from 2011 to 2023. With data from UNHCR, 3RP, and global reports, I charted migration routes, host countries, and humanitarian needs, transforming raw numbers into a story of human resilience.

QGIS illuminated the geography of displacement, mapping origin cities, transit routes, and major refugee camps across Europe and the Middle East. Three thematic maps highlight key dimensions: one traces primary migration paths and camps, another spotlights top host nations such as Turkey and Germany, and a third reveals the border fences that have shaped refugee flows.

Flourish brought these insights to life through interactive data visualizations: a line graph tracks external and internal displacement, a pie chart displays top humanitarian funders, a bar chart explores refugees’ main challenges, and a population pyramid exposes stark demographics—where children often outnumber adults.

This project blends rigorous data cleaning and geospatial analysis with empathetic storytelling, illuminating the Syrian refugee crisis’s immense scale and profound human toll. Despite fragmented datasets and inconsistent formats, I meticulously cleaned and harmonized data in Excel, ensuring accuracy for QGIS’s detailed spatial mapping and Flourish’s interactive visualizations. By merging these tools, it offers a dual lens—analytical precision to uncover migration patterns, host countries, and demographic trends, alongside a heartfelt narrative that captures the personal journeys behind the numbers.

GET IN TOUCH

Interested in my work or want to collaborate?Feel free to reach out using the form below 👇

Contact

• My CV

• LinkedIn
• Fiverr
• Freelancer
• Behance
• Dribbble
• Email