Solid ideas that make students pay attention.

---

1. “FoodFresh” – Predict Whether Vegetables Are Fresh or Spoiled (Image ML)

Why it works

Food is universal. Students instantly get it — no deep theory needed.

Core idea

Use a tiny image dataset (fresh vs spoiled tomatoes, potatoes, apples).
Train a small CNN (or MobileNet) to classify fresh / rotten.

Class impact

Shows how machines “see”, a topic students always find magical.

---

2. “SpamGuard” – Email/SMS Spam Detection

Why it works

Everyone hates spam. Students relate instantly.

Core idea

• Use SMS Spam Collection dataset
• TF-IDF + Naive Bayes
• Predict spam or not spam

Why students love it

Because seeing ML block fake offers and scams feels real.

---

3. “SleepStyle” – Predict Sleep Quality from Daily Habits

Why it works

Health + ML = high curiosity.
And it’s simple, not bloated.

Data

Synthetic dataset with:

• hours_sleep
• screen_time
• steps
• caffeine_intake
• mood

Prediction: Good Sleep / Bad Sleep

Why it clicks

Students see ML applied in everyday life — not in some far-off AI lab.

---

4. “TrafficSense” – Predict Traffic Jam Level from Time & Weather

Core idea

Small dataset with:

• time_of_day
• day_type (weekday/weekend)
• rainfall
• temperature
• accidents_reported

Predict: Low / Medium / High Traffic

Strong teaching point

Shows traditional ML mindset:
feature → relation → pattern → prediction.

---

5. “FakeTweet Detector” – Identify Fake News Tweets

Core idea

Scrape 40–50 tweets (real + fake).
Vectorize → train classifier → predict truth value.

Why it grabs attention

Social media is their world — they perk up when ML enters it.

---

6. “PriceSense” – Predict House Price from 5 Features

Why it works

This is the classic ML rite of passage.
And classic things hold weight — they survive trends.

Core idea

• bedrooms
• area
• location_encoding
• age
• parking

Predict price (Regression).

---

7. “StyleMatch” – Outfit Color Recommendation using Clustering

Why it works

Fashion + ML = instant engagement, especially for beginners.

Core idea

• Extract dominant colors from shirts (using K-Means)
• Recommend matching pant colors

Teaching impact

Students see unsupervised learning come alive visually.

---

8. “StudyBoost” – Predict Student Performance from Study Habits

Data

• hours_studied
• attendance
• assignments_completed
• distractions
• sleep_hours

Predict: Pass / Fail / High Score

Why students care

Because the model is indirectly talking about them.

---

9. “EmotionBeats” – Predict Mood from Music Tempo & Genre

Core idea

Small dataset with:

• bpm
• energy
• genre
• key
• loudness

Predict: Calm / Energetic / Sad / Happy

Why unforgettable

Music catches hearts before minds — perfect teaching tool.

---

10. “LoanRisk MiniChecker” – Predict Loan Approval Risk

Core idea

• income
• age
• loan_amount
• job_type
• previous_defaults

Predict risk: Low / Medium / High

Why it matters

Shows how ML shapes finance and real decisions.