AI Fundamentals & Prompting
Data Quality Matters
Example: Email Spam Filter (Old Way)
IF email contains "FREE MONEY" → Mark as spam
IF email contains "CLICK HERE NOW" → Mark as spam
IF sender is unknown → Mark as spam
Problem: You need rules for EVERY scenario. Spammers evolve faster than you can code.
Example: Email Spam Filter (ML Way)
STEP 1: Feed 1 million emails (500K spam, 500K legit)
STEP 2: Machine learns patterns that predict spam
STEP 3: System predicts on NEW emails
Result: Gmail catches 99.9% of spam. Machine adapts as it sees more examples.
Spam or Not Spam?
Will Churn or Stay?
Approve or Deny?
Disease Present?
What will it sell for?
How much will they spend?
What will price be?
How many units sell?
Question: Will this user watch this show? Features: Watch history, ratings, time of day
Question: What should we price this item at? Features: Cost, demand, competitor prices
Question: What is this object? (Pedestrian, car, sign?) Features: Camera images, distance, speed
83%
93%
1. Previous Purchases — Most important (past behavior predicts future)
2. Time on Website — Second most important (engagement signals intent)
3. Income — Third most important (ability to buy)
4. Age — Least important (not a strong predictor)
This tells business strategy: Focus on engagement and repeat behavior!
Dataset: 2,000 telecom customers
Which customers will cancel service next month?
Train a model, document accuracy, identify top factors causing churn
Dataset: 5,000 used cars
What should this used car sell for? (Fair market price)
Train a model, document R² score, test prediction on sample car
% of correct predictions
Target: 75%+
How much variation model explains (0-1)
Target: 0.70+
Setup & Log In
Exercise 1: Churn Model
Exercise 2: Price Model
I'm here to help anytime
FirstName_LastName_ChurnModel.pdf
FirstName_LastName_PriceModel.pdf
Same tools Netflix & Amazon use
You identified what drives outcomes
For your resume & LinkedIn
Not theoretical—you did it
✓ You trained a churn prediction model (Classification) with 75%+ accuracy
✓ You trained a price prediction model (Regression) with R² 0.70+
✓ You identified top features that predict outcomes
✓ You tested a prediction on new data
✓ You understand the difference between Classification & Regression