How I stumbled into data analytics, survived messy Excel files, and helped visualize the state of education across the Philippines.
I didn't start this project as a data analyst.
I was a fullstack developer — comfortable in my lane, writing APIs, building UIs, thinking in components and endpoints. Then the project landed on my desk: build a data analytics dashboard for the Department of Education of the Philippines. Visualize learner enrollment, school data, and personnel resources across every region in the country.
I said yes. I had no idea what I was getting into.
The Problem: A Goldmine of Data, No Way to See It
DepEd sits on top of one of the most important datasets in the country. We're talking about millions of learners across thousands of schools — from Batanes to Tawi-Tawi. But none of it was consolidated in a way that made it usable for decision-making.
The data covered everything:
- Learners — enrollment by grade level, level of education, sector, SHS track, Alternative Delivery Mode (ADM), ALIVE program, and IP (Indigenous Peoples) communities
- Schools — formal vs. ALS (Alternative Learning System), sector, subclassification, curricular offerings
- School Personnel and Resources — number of teachers, classrooms, seats, toilets, water supply, internet connectivity, and more
The ask was to build dashboards that would let DepEd management see all of this at a glance — national overviews down to per-school drill-throughs.
Simple enough, right?
Before the Challenges: I Wasn't Alone
I want to say this upfront before getting into the hard parts: I didn't do this by myself.
I had a colleague who was on the same unfamiliar ground as me at the start. Neither of us were data people — we actually took the Power BI certification together, studying side by side, fumbling through the same DAX errors, building things wrong and rebuilding them. Over time, they grew deep into the data domain in a way that was genuinely impressive to watch. By the later stages of the project, they had become the kind of person you'd actually call a data expert — and having that person to think through problems with made all the difference.
Our company also gave us the room to experiment, fail, and figure things out without the pressure of getting it perfect on the first try. That kind of support is rarer than it should be, and it was a big reason the project survived to the finish line.
Okay. Now the challenges.
Challenge #1: The Data Was Behind a Wall
The first roadblock hit early: DepEd's data lived in a SQL database they were very reluctant to share. Understandably so — this is sensitive government data. But it meant we couldn't connect directly to the source.
Their solution? Share Excel summaries.
That became our entire data pipeline. An Excel file. Uploaded to SharePoint. That was the main data source for a national-scale dashboard.
If you've ever tried to build a reliable data model on top of Excel files generated by humans, you already know the specific kind of pain that followed.
Challenge #2: Learning a New Craft Mid-Flight
Here's the part I didn't anticipate: I had to become a different kind of developer — while the project was already moving.
Power BI wasn't something I had seriously worked with before. Neither was the broader discipline of data modeling, DAX, or Power Query. So I went and got my Microsoft Certified: Associate Data Analyst certification — not as a nice-to-have, but as a survival mechanism. My colleague did the same, and we pushed each other through it.
A lot of what I learned, I learned by doing it wrong first. Experimenting at midnight. Breaking a data model and spending hours tracing why a single measure was returning the wrong number for one region but not another. It was the kind of frustration that quietly builds competence.
Challenge #3: No Consolidated Vision from Stakeholders
One of the quieter struggles of the project — and honestly one of the most common in any enterprise project — was that DepEd management didn't have a unified picture of what they wanted.
Different stakeholders had different ideas. What metrics mattered? What comparisons were meaningful? What story did they want the dashboard to tell?
This led to multiple rounds of iteration — redefining data definitions mid-build, reworking visualizations, re-scoping what "enrollment" even meant in a given context (gross enrollment? net? by sector?). Every iteration taught me something new, but it also meant nothing was ever truly "done."
Challenge #4: Data Modeling at Scale is Hard
The part that kept me up at night was the data modeling.
The DepEd data has multiple levels of hierarchy — national → regional → division → district → school. And then within schools, you have sub-layers by grade level, track, sector, program, and so on. Modeling relationships across Excel sheets that weren't designed with a schema in mind was an exercise in patience.
Power Query helped us transform the data, but we kept hitting redundancies — the same dimension appearing in multiple fact tables, cardinality mismatches, relationships that would work for one visual but silently break another. There was a point where we had a measure that looked correct on the national overview but was double-counting enrollees at the regional level because of how we'd joined two sheets with overlapping hierarchies. Tracking that down took days. We had to rebuild our data model more than once.
By the end, I had a genuine respect for people who do data engineering full-time. It's work that looks invisible when done right — and brutally obvious when done wrong.
The Stack — and Why It Changed
We started with Power BI and eventually migrated to Microsoft Fabric about midway through the project. The move wasn't just about scale — it was about sustainability. As the number of reports grew and the data transformations became more complex, we needed a proper analytics environment rather than a collection of .pbix files pointed at a SharePoint folder.
Fabric gave us a unified workspace: lakehouses for data storage, dataflows for transformations, and a shared semantic model that multiple reports could draw from consistently. It also made collaboration significantly easier — no more emailing model files back and forth.
The SharePoint-hosted Excel remained our source of truth throughout. Not ideal, but we built defensively around it — validating totals, flagging anomalies, and maintaining a transformation layer that could absorb changes in the upstream file structure without breaking every report downstream.
Two Years Later
This project ran for about two years. It started with me Googling what DAX stood for and ended with me designing multi-layered semantic models and presenting dashboards to DepEd leadership.
I think about what this project represents: millions of Filipino learners, captured in data — their schools, their teachers, the classrooms they sit in, whether there's internet, whether there's a clean toilet. That data, finally visible. Finally consolidated. Finally useful.
That's worth a lot of late nights and broken data models.
What I'd Tell Anyone Starting Something Like This
- Say yes before you're ready. The certification, the skills, the confidence — those came because I committed, not before.
- Find your person. A teammate who's learning alongside you — equally lost, equally invested — is worth more than any course or tutorial. Build things wrong together. You'll figure it out faster.
- Stakeholder alignment is a technical problem. Ambiguous requirements will break your data model. Invest time upfront to nail down definitions.
- Excel as a data source is a constraint, not a dealbreaker. Work with what you have, and build defensively around it.
- Data modeling is a skill, not a task. Treat it like architecture. Get it wrong and everything downstream suffers quietly — until it doesn't.
- Impact matters. When the work is for something real — education, public service, people — it hits differently. Let that keep you going on the hard days.
Thanks for reading. I'm Christian Bandibas, a developer based in the Philippines. If you've worked on public sector tech or data projects, I'd love to hear your experience in the comments.
Tags: #dataanalytics #powerbi #microsoftfabric #philippines #publicsector #careergrowth #dataengineering
Top comments (0)