Lesson 11: Module 7 Hands-on Projects

In this final module project, you will apply everything you've learned about the Scikit-Learn pattern to solve real-world problems. Choose one of the three paths below.

---

Project 1: The Luxury Home Predictor (Regression)

Objective: Build a model that predicts house prices based on multiple features.

• Task:
  1. Create a dataset with SqFt, Bedrooms, and Age_of_House.
  2. Use LinearRegression to train the model.
  3. Predict the price of a 10-year-old, 3000 sq ft house with 4 bedrooms.
  4. Evaluate using Mean Squared Error (Self-research tip: from sklearn.metrics import mean_squared_error).

Project 2: The E-commerce Segmenter (Unsupervised)

Objective: Group customers based on their spending habits.

• Task:
  1. Create a dataset with Annual_Income and Spending_Score.
  2. Import KMeans from sklearn.cluster.
  3. "Fit" the model to find 3 natural groups of customers.
  4. Visualize the clusters using a Seaborn Scatter Plot where hue is the cluster center.

Project 3: Social Media Sentiment Analyzer (Classification)

Objective: Detect if a tweet is "Positive" or "Negative."

• Task:
  1. Create a list of 20 sample sentences (half positive, half negative).
  2. Use the make_pipeline method with CountVectorizer and MultinomialNB.
  3. Train the model.
  4. Test it with a new sentence: "This is the worst experience I've ever had."
  5. Print the Classification Report (Self-research tip: from sklearn.metrics import classification_report).

---

Module 7 Recap: Exercises and Quiz

Exercise 1: The Metric Matcher

Match the metric to its use case:

1. Precision
2. Recall
3. Accuracy

A. You want to find EVERY possible case of fraud, even if you have some false alarms. B. You want to ensure that every "Yes" guess is absolutely certain. C. You have an equal number of apples and oranges and want to know how many you got right.

Exercise 2: The Model Swapper

Take your code from Project 1. Replace LinearRegression with RandomForestRegressor. Does the prediction change? Which one do you trust more?

---

Module 7 Quiz

1. What is the Variable Name convention for the "features" (inputs) in Scikit-Learn? A) y B) x C) X D) target

2. Which algorithm is best for predicting a continuous number? A) Logistic Regression B) Decision Tree Classifier C) Linear Regression D) Naive Bayes

3. What does "Overfitting" mean? A) The model isn't smart enough. B) The model has memorized the training data and can't handle new data. C) The data is too small to use. D) The computer ran out of memory.

4. Why is a Random Forest usually better than a single Decision Tree? A) It's faster to train. B) It uses less memory. C) it combines the votes of many trees to improve stability. D) It doesn't require any math.

5. Which metric is most important for a doctor trying to detect a deadly disease? A) Precision B) Accuracy C) Recall D) F1-Score

---

Quiz Answers

1. C | 2. C | 3. B | 4. C | 5. C

---

Final Course Conclusion

You have finished the "Python from Basics to AI" course. You have the foundations of programming, the structural skills of OOP, the data mastery of Pandas/NumPy, and the predictive power of Scikit-Learn.

The world of AI is now open to you. Go forth and build!