Species (K-Means) Cluster Analysis: A Case Study on the Limitations
This exploratory project applies unsupervised clustering (K-Means) to the classic Iris dataset to investigate whether iris species can be segmented based on petal and sepal measurements. Originally popularized by Ronald Fisher in 1936, the dataset includes 150 samples across three species—Iris setosa, Iris versicolor, and Iris virginica. While the data and species distinctions are well-studied, this analysis serves as a hands-on case study in clustering. The aim was to highlight both its potential and its limitations, particularly in contexts where true class boundaries may not align with cluster groupings. The project concludes with a reflection on why supervised approaches like logistic regression are more suitable when class labels are known.
Optimal Market for Advertising E-Learning Products
This project aims to identify the most promising markets for an e-learning company with online courses in web development, mobile development, data science, and others. Through exploratory data analysis, I evaluate trends in programming interest, learner demographics, and purchasing potential to recommend two to three optimal regions for advertising investment. The analysis is guided by key criteria: geographical distribution of new coders, willingness to pay for a monthly subscription, alignment of interests with the company’s offerings, English language accessibility, and insights into learning habits and demographic factors.
Investigating Bias in Online Movie Ratings
This project explores potential biases and inflation in online movie rating systems, focusing on how these factors may influence consumer trust and decision-making. Inspired by data journalism on movie rating manipulations, I analyze historical and recent movie ratings to evaluate the prevalence of systemic rounding practices and inflated scores. Key concerns include the near absence of low ratings, a disproportionate concentration of scores above 4 stars, and the potential for commercial platforms to benefit from higher perceived movie quality. Given the film industry’s substantial economic impact and the influence online ratings wield over audience choices, evaluating the integrity of these systems is essential. The analysis aims to shed light on how these biases may persist or have evolved over time, ultimately aiming to contribute to a transparent and trustworthy rating systems for consumers.
Excel Ad-hoc
Statistical Insights into Vehicle Fuel Efficiency
This project explores the factors influencing vehicle fuel efficiency (MPG) using statistical analysis and visualization in Excel. A two-sample t-test was conducted to compare the average MPG between cars manufactured in Europe and the USA, revealing a statistically significant difference favoring European vehicles. Additionally, linear regression models highlighted relationships between MPG and two key predictors: vehicle weight and horsepower. The analyses showed strong negative correlations, with approximately 69% of MPG variation explained by weight and 60% by horsepower, highlighting their predictive power in understanding fuel efficiency trends.
Hypothesis Testing: Z-Test for Data Scientist Salaries
Conducted a two-tailed Z-test in Excel to verify if the average data scientist salary significantly differs from a reported value of $113,000. The goal was to formulate null and alternative hypotheses and determine if the true mean salary significantly differs from the reported figure using a two-tailed Z-test at various significance levels (0.1%, 0.5%, 5%, 10%).
Manufacturing Line Productivity Analysis
This Excel-based project analyzes production efficiency and operator performance across a beverage manufacturing line. It integrates batch-level timing data, product specifications, and downtime factors to assess productivity. Key metrics such as efficiency ratios, top-performing operators, and root causes for downtime are calculated and visualized through a structured data model. A Pareto analysis and a dashboard view highlights the most significant delays, offering actionable insights to improve operational performance.
Retail Sales Dashboard & Exploratory Analysis
This Excel project involved end-to-end exploratory data analysis, starting with data cleaning and preparation using custom calculations. Insights were highlighted through conditional formatting and data bars, followed by the creation of an interactive dashboard. The dashboard highlights monthly revenue trends, transaction patterns by weekend and hour, and product performance, including detailed views by category. All visuals are dynamically filterable across store locations for easy exploration.
Power BI & Tableau
Interactive Financial Sales Dashboard
This is an interactive financial sales report in Power BI with visually dynamic charts and drill-through navigation. The dashboard presents high-level insights such as sales performance by country, product, and segment, as well as quarterly trends for both sales and cost of goods sold (COGS). Users can explore detailed data tables through paginated buttons and filter views to focus on specific regions like North America or Europe. Monthly sales trends are also visualized by country and product category, enabling intuitive and strategic data exploration.
SQL Ad-Hoc Analysis
HR & Business Insights with SQL
Ad-hoc analysis to support HR, marketing, and finance departments by creating SQL views for dashboard-ready insights. Features analytical queries to:
• Track current salaries by department and role using joins and date filters
• Segment customer behavior based on rental return patterns using CASE and view chaining
• Support valuation analysis by aggregating district-level rental and customer data
• Prepare HR-friendly queries to assess manager salaries and departmental salary averages
Demonstrated proficiency in SQL views, data transformation, and business-focused analytics within a relational database.
JavaScript Web Development
Tracalorie App
Coming soon.
Flixx Movie App
Coming soon.
Random Ideas App
Coming soon.