Most companies don’t fail at AI because of models. They fail because of data.
More specifically:
Continue readingThey try to run AI workloads on top of a data warehouse that was never designed for it.
Can We A/B Test an LLM? A Practical Guide to LLM Experimentation
As large language models (LLMs) move from demos into production systems, one question comes up quickly:
Can we A/B test an LLM like we test product features?
For teams coming from product analytics or growth, the instinct is straightforward:
- Build variant A
- Build variant B
- Run an A/B test
- Measure impact
Commonly used python packages for data analysts
If you’re working as a data analyst today, Python is no longer just a “nice to have.” It’s part of the daily toolkit.
But here’s the reality:
Continue readingMost analysts don’t need dozens of libraries — they need a small, reliable stack, used well.
Will BI Tools Survive the AI Era
Over the past two years, artificial intelligence has rapidly entered the world of data and analytics. Large language models can now write SQL, explain trends, generate dashboards, and summarize reports in seconds. Because of this, a common question is starting to appear in data teams and executive meetings:
Continue reading
AI Won’t Replace Analysts — But It Will Replace Shallow Analysis
AI is getting uncomfortably good at the parts of analytics that used to look impressive.
It can:
- write SQL from a sentence
- summarize a dashboard in seconds
- generate charts and slide narratives
- explain metric movements in fluent English
So it’s natural to wonder: Will AI replace analysts?
Continue reading
From BI to AI: What Actually Changes (and What Doesn’t)
Over the last few years, “AI transformation” has become a board-level conversation.
Executives ask:
- Are we still doing BI?
- Should dashboards be replaced by copilots?
- Do we need fewer analysts and more machine learning engineers?
Underneath the noise, a more important question exists:
Continue readingWhat actually changes when an organization moves from BI to AI?
How Tree-Based Models Actually Create (or Fail to Create) Business Value
Tree models are among the most widely used machine learning methods in modern business systems.
From fraud detection and churn prediction to logistics risk scoring and pricing optimization, tree-based models power decisions across industries.
Continue reading
From SQL Experiment Analysis to Causal Inference
Most analysts start experiment analysis in SQL.
You write a query.
You compare treatment vs control.
You calculate lift.
You declare a winner.
For many well-designed A/B tests, that’s perfectly valid.
Continue reading
When Tree Models Actually Beat Logistic Regression (and When They Don’t)
If you work in applied data science long enough, you’ll eventually hear some version of this question:
“Should we switch from logistic regression to a tree-based model?”
Sometimes the answer is yes.
Very often, the answer is not really.
Logistic Regression Is Still King: Here’s Why
In a world filled with gradient boosting, deep learning, and AutoML tools, logistic regression can feel almost embarrassing to mention.
It’s old.
It’s simple.
It’s taught in every intro course.