How I Test My Data Skills Before Every Interview — 5 Self-Assessments That Showed Me What I Didn't Know

#datascience #career #sql #python

Six months ago I started a self-directed data science program. Every day I study something — SQL joins, pandas groupby, scikit-learn pipelines, hypothesis testing, Power BI measures.

But after months of learning, I hit a wall: I didn't know what I didn't know.

I could follow tutorials fine. I could explain concepts when someone asked. But put me in front of a blank editor with a question I hadn't seen before? That's when the gaps showed up.

So I built a system. Five mini-assessments — one for each core data skill. Here's what I learned.

The 5 Skill Areas That Matter Most

Looking at hundreds of data analyst job descriptions, the same 5 skills kept appearing:

SQL — Everyone asks for it. Not just SELECT * FROM, but window functions, CTEs, and performance.
Python/Pandas — Data wrangling is 80% of the job. Merges, groupby, apply, handling missing data.
Statistics — Hypothesis tests, p-values, confidence intervals. You need to know when to use which test.
Machine Learning — Not just fit/predict, but pipelines, cross-validation, and knowing when a model is actually good.
BI & Visualization — Power BI, Tableau, and the ability to communicate insights visually.

How Each Assessment Works

I didn't want all multiple choice (too easy to guess) or all coding (too time-consuming). So each assessment mixes three formats:

MCQ (1 pt each) — Quick knowledge checks. Do you know what HAVING does?
Short Answer (2 pts each) — Can you explain the difference between RANK and DENSE_RANK in your own words?
Practical (3 pts each) — Write the actual query or code. This is where gaps really show up.

What I Discovered

The first time I ran through all 5 assessments, my scores were uneven:

SQL: Strong (I'd been practicing daily)
Python: Intermediate (could wrangle but couldn't optimize)
Statistics: Beginner (p-hacking was a real risk)
ML: Beginner (knew the theory, couldn't build a pipeline from scratch)
BI: Intermediate (Power BI was fine, DAX was weak)

This told me exactly where to focus. No more wasting time on skills I already had. No more guessing.

The Scoring System

I used a simple 3-level system:

Try It Yourself

I packaged these 5 assessments into a downloadable PDF with full answer keys, score trackers, and a personalized roadmap. If you're preparing for data interviews or just want an honest self-assessment, it might help you too.

I'm Deepanshu — self-taught data scientist, building tools for data learners.