As a mentor to Kerala PSC aspirants, I’ve always believed that consistency is the ultimate game-changer in competitive exam preparation. Talent matters, but what separates top performers is not brilliance — it’s the ability to show up and learn every single day.
But the question many students ask is:
“How do I develop a consistent learning habit?”
🌱 Introducing Students to Habit Science
To help my students build better learning habits, I started teaching them the principles from:
- Atomic Habits by James Clear
- The Power of Habit by Charles Duhigg
We explored the habit loop:
👉 Cue → Routine → Reward
Understanding this helped my students realize that learning is not about motivation, but about systems.
📊 Building a System: My Habit Tracker Experiment
To put this into practice, I created a simple yet powerful system:
- Students filled Google Forms multiple times a day to report their study times.
- The data went into Google Sheets, visualized through basic dashboards.
- I used Google Apps Script to build automated triggers and summaries.
🚀 The impact was instant:
- Students said the form acted as a trigger, nudging them to study.
- Many increased their total study time weekly.
- Some even reported improved focus and planning.
This was my first real experiment combining mentoring + behavioral science + tech.
🤔 Then I Thought Deeper… Can I Learn From Their Learning?
Looking at all this data, a question hit me:
“What if I could identify learning patterns in their habits? Could I give better, more personalized feedback?”
This is where I stumbled into the world of Machine Learning — specifically, Supervised and Unsupervised Learning.
📘 What is Supervised Learning?
Supervised Learning is when a model learns from labeled data — meaning, for every input, the correct answer is already known.
- Think of it like teaching a child using a solved question paper.
- The model studies the examples and learns to map inputs to the right outputs.
🧪 My Example:
Predicting a student’s grade based on their percentage:
textCopyEditIf score < 40% → Grade D
If score 40–50% → Grade C
If score ≥ 50% → Grade B
Here, the model sees many such input-output pairs and learns how to grade new scores.
📘 What is Unsupervised Learning?
Unsupervised Learning is when the data has no labels — the model isn’t told the correct answers. Instead, it looks for hidden patterns or natural groupings.
It’s like giving students blank answer sheets and asking the model to find out who studied similarly, just by comparing their behavior.
🧪 My Application:
I looked at how students reported their study behavior:
- Time of day
- Duration of study
- Frequency
No labels like “good learner” or “struggling student” — just raw behavior data.
Using clustering techniques, I could group students into learning patterns, like:
- Consistent daily learners
- Weekend-only studiers
- Night owls who study in bursts
🔁 Summary Table
| Feature | Supervised Learning | Unsupervised Learning |
|---|---|---|
| Data | Labeled (input + correct output) | Unlabeled (no answers) |
| Goal | Predict/classify future outcomes | Discover hidden patterns or clusters |
| My Use Case | Predicting grades or pass/fail | Grouping students by habit patterns |
| Analogy | A teacher with answer keys | A detective finding clues |
🤖 Now I’m Building an AI-Powered Mentoring Tool
All of this has led me to my next step:
“Can I build an AI system that tracks student habits, learns patterns, and gives intelligent feedback?”
I’ve already started working on:
- Automating data collection and clustering
- Using AI to detect habit types
- Giving customized study advice through a learning assistant
I’m currently exploring GPT-based models, supervised classifiers, and unsupervised clustering algorithms to take this idea to the next level.
✍️ Final Thoughts
This journey started with a simple question: “How do I make my students consistent?”
It led me into behavior science, systems thinking, and now, applied AI.
If you’re an educator, mentor, or learner:
Start small. Track habits. See patterns.
You may not just change your study sessions — you might build the next AI mentor.
Stay tuned — I’ll soon share how I’m training my own AI to do all of this 🔥