Posts

Background As a data analyst on critical projects, I optimize PostgreSQL queries to ensure end users receive accurate data because even a single second can make a difference in time-sensitive analyses. In resource-constrained environments, cloud computing costs can also limit access to essential tools, slowing decision-making, reducing team transparency, and even threatening a project’s success. That’s why SQL performance tuning matters. Over the past year, I’ve gained substantial experience in query optimization, yet I was surprised by how scarce practical resources were when I was actively tackling these challenges. Many nuances were obscure, and some even eluded more senior colleagues. ...

Background A small thing I’ve been thinking more about lately: git commit messages. In analytics work, commit messages aren’t just about code; they’re about documenting how business logic evolves. A small change to a SQL query can quietly change KPIs, forecasts, or stakeholder decisions. Without clear commits, that history is lost. In analytics, commits often capture changes that won’t surface in code reviews or dashboards immediately. That makes commit history one of the few reliable ways to reconstruct how metrics evolved over time. ...

Let’s be real—data visualization can be frustrating, even for power analysts. I often get so buried in the numbers that stepping back to cOmMuNiCaTe insights can feel like a chore. It’s like making a delicious meal but skipping the plating; no matter how good it tastes, it won’t impress if it looks like mush. But visuals matter. A lot. The good news? You don’t have to sacrifice hours to create dashboards that are sharp, clear, and beautiful. Here are five go-to tips to level up your visualizations—without burning out. The tips are tailored to Tableau users but can be repurposed for your preferred data visualization tool. ...

Background Most dashboards don’t fail because of bad analysis; they fail because they’re presented poorly. I’ve seen presentations where every chart was explained in detail, and the audience still walked away without a clear takeaway. I took Andy Kriebel’s Makeover Monday challenge a step further by thinking about something I don’t see discussed often: how to actually present a dashboard. Especially when it’s a one-page overview, the challenge isn’t building it—it’s walking someone through it in a way that lands. ...

Background Most data analysts learn Python in a very procedural way: write a script, transform some data, export a result, repeat. That works. Until it doesn’t. As projects grow, logic gets duplicated, scripts become harder to maintain, and simple changes start breaking multiple parts of your workflow. That’s usually the point where OOP (Object-Oriented Programming) starts to matter. My mentor once told me to use Python like a developer—not just as an analyst. That stuck with me. Understanding data structures is important, but learning how to structure code is what actually makes your work scalable. ...

Background During the early months of COVID-19, job losses were widely reported—but I wanted to understand how those impacts varied across sectors and provinces, and how quickly recovery actually began. Specifically: Which sectors were hit hardest? Did recovery happen uniformly across provinces? Were there meaningful differences across demographic or regional patterns? To answer this, I analyzed labor market data across Canadian provinces from January to September 2020 using R. Approach I worked with multiple datasets covering employment by province, sector (NAICS categories), and month. The workflow was straightforward but important: ...

Background For a political advocacy organization, I analyzed web traffic and conversion data to understand what drives user engagement, which segments generate the most revenue, and where optimization opportunities exist. The goal was straightforward: turn raw web activity into actionable insights for the web and fundraising teams. Approach I worked with two core datasets—page_views for user behavior and conversions for donations—focusing on traffic patterns, device segmentation, referral sources, and donor value. ...

Cohort tables offer valuable insights into customer behavior by grouping them based on shared attributes, in this case, acquisition date or date of first purchase. They are fantastic for identifying whether you have a retention problem or a user acquisition problem and can allow marketers to understand user behavior through experiments. Learning to read these tables can be tricky at first but effectively empowers you to analyze customer engagement, retention, and acquisition trends. ...