Parthiban Kannan

Home
India
Cuddalore
Parthiban Kannan

🚀 Inspiring Students to Explore Technology 💻✨
📊 Data Analytics | 📈 Business Intelligence | 📱 App Development
👇🏻 Learn with **Parthiban Kannan** 💡🎯

17/05/2026

It make sense 🥺

12/05/2026

Unpopular Opinion 👀
Most companies won’t pay you more just because you know 10 frameworks.

They pay more when you can:
✅ Explain clearly
✅ Present confidently
✅ Handle clients & teams
✅ Convert ideas into business value
Technical skills may get you the interview…
Communication skills often decide your salary. 💯

What do you think? Agree or disagree? 👇

11/05/2026

This is how Claude & ChatGPT thank each other when I switch between the tools.

10/05/2026

Did I miss any other need for Vibe Coders? 😆🤣

27/03/2026

Seriously, for beginners who think they’ve mastered Power BI in just a couple of hours through an online course.

If you don’t know DAX, you don’t know Power BI. End of story.

Follow Parthiban K. for AI & Analytics Insights 💡Seriously, for beginners who think they’ve mastered Power BI in just a couple of hours through an online course.

If you don’t know DAX, you don’t know Power BI. End of story.

Follow Parthiban K. for AI & Analytics Insights 💡

26/03/2026

Key Areas of Data Cleaning Using SQL Every Data Professional Must Know.

1. Data Ingestion & Exploration 📥🔍
This initial stage involves loading raw data into SQL tables and using SELECT * or COUNT() queries to understand the schema. It helps identify the scale of the dataset and provides a baseline for the cleaning tasks ahead.

2. Identify Missing Values 🕳️⚠️
SQL uses IS NULL filters and COUNT(CASE WHEN...) logic to locate gaps in the data where information is absent. Once identified, you can either drop these rows or use COALESCE to fill them with default values or averages.

3. Duplicate Removal 🧬❌
Redundant records are detected using GROUP BY and HAVING COUNT(*) > 1 on unique identifiers like Order IDs. You can then use Common Table Expressions (CTEs) with ROW_NUMBER() to partition the data and delete the extra copies.

4. Data Type Conversion 🔄📊
This step ensures consistency by using the CAST or CONVERT functions to change strings into dates or integers. Proper data types are critical for performing accurate mathematical calculations and time-series analysis later on.

5. Formatting & Normalization 🧹🔤
SQL functions like TRIM() remove leading spaces, while LOWER() or UPPER() standardize text case across the board. This prevents “New York” and “new york” from being treated as different entries during data aggregation.

6. Handle Outliers & Consistency 🚨✅
The final stage uses WHERE clauses to filter out statistically impossible values, such as negative prices or future birth dates. This ensures the remaining data is logical, consistent, and ready for high-level AI and analytics modeling.

Save this for Later.

25/03/2026

🚀 Stop trying to learn EVERYTHING to get into LLMs

LLMs are NOT about starting from scratch ❌
They’re about focusing on what actually matters at each stage 🎯

Here’s the real game 👇

1️⃣ Foundation
🤔 What people actually think: You need hardcore math
✅ But what actually it is: Python + basics + consistency

2️⃣ Model Architecture
🤔 What people actually think: Build Transformers from zero
✅ But what actually it is: Just understand how they work

3️⃣ Pre-Training
🤔 What people actually think: Train models with huge GPUs
✅ But what actually it is: Use pre-trained models smartly

4️⃣ Fine Tuning & Adoption
🤔 What people actually think: Fine-tuning is mandatory
✅ But what actually it is: Prompting + RAG does most of the job

5️⃣ LLM Ops & Deployment
🤔 What people actually think: Just call an API
✅ But what actually it is: Scale, latency & cost matter

6️⃣ Advanced Concepts
🤔 What people actually think: Learn everything in AI
✅ But what actually it is: Go deep in what truly matters

🔥 Focus > Overwhelm
🔥 Depth > Noise

05/03/2026

You can’t compare Machine Learning projects with traditional software projects. ⚙️🤖

In software projects, you can deploy the product and wait for customer feedback to add new features or modify existing ones. 🧩

But that approach doesn’t work in Machine Learning.

If you deploy a model and simply wait over time, the result won’t be feedback — it could become a catastrophic product failure. ⚠️

Because ML models depend on data patterns, and when those patterns change, the model’s predictions start degrading. 📉

Without monitoring and retraining, the system quietly starts making wrong decisions. 🔁🧠

27/02/2026

You’ve delivered 50 dashboards. 📊
But someone else is still making the calls. 🎯

That’s not an experience problem.
That’s a thinking problem. 🧠

Tool thinking keeps you executing. 🔧
System thinking makes you the expert in the room. ⚡

Save this. You’ll need it. 👇

19/02/2026

Most ML models don’t fail because they’re bad. They fail because they solve the wrong problem. 🚫

Anomaly detection says:
“Something is wrong.”

Predictive maintenance says:
“This machine will fail in 48 hours. Schedule maintenance.”

See the difference?
Professionals build accurate models.
Experts map ML to business decisions.

Before choosing an algorithm, ask:
👉 What decision is the business trying to make?

That’s how you move from coder to strategist. 🚀

14/02/2026

The most dangerous stakeholder isn’t the one who knows nothing about ML.

It’s the one who took a 2-week online course. 🚩

Here’s what happens:

They expect 90% accuracy on day 1.

They set model metrics as success criteria.
They don’t understand why “accurate” models fail in production.And they’re convinced they know better because they “learned ML.”I’ve seen this pattern destroy projects over my 14 years in the industry.

Here’s the reality they miss:
❌ Real-world data is messy—not like clean Kaggle datasets
❌ Training accuracy ≠ Production performance
❌ Even 90% accurate models can fail in production
❌ Business impact > Model metrics

So what should you do instead?
✅ Set BUSINESS METRICS, not just model metrics
✅ Understand the cost of wrong predictions (False Positives vs False Negatives)
✅ Define acceptable error rates based on business impact
✅ Reset stakeholder expectations EARLY in the project

The biggest project killers aren’t technical problems.They’re stakeholders with partial knowledge setting unrealistic expectations.

→ The exact red flags to spot early
→ How to handle these conversations like a senior professional
→ The business metrics framework that actually works

Don’t let partially-informed stakeholders kill your next project.

To become an expert in AI & Analytics, follow for real-world frameworks—not theory.

Drop a 🚩 in the comments if you’ve experienced this!

Address

Cuddalore
607001

Website

http://www.theparthee.com/

Alerts

Be the first to know and let us send you an email when Parthiban Kannan posts news and promotions. Your email address will not be used for any other purpose, and you can unsubscribe at any time.

Want your school to be the top-listed School/college in Cuddalore?