Practice: Dataset Preparation & Validation

# Practice: Build a simple grade predictor using the full AI cycle
# Step 1: Data, Step 2: Learn, Step 3: Predict, Step 4: Evaluate

import random

# Step 1: Generate training data
# Features: [study_hours, sleep_hours_before_exam]
# Label: exam_grade (A, B, C, D)

def generate_grade(study, sleep):
    score = study * 8 + sleep * 4
    if score >= 80:
        return "A"
    elif score >= 60:
        return "B"
    elif score >= 40:
        return "C"
    else:
        return "D"

# Create training dataset
all_data = []
random.seed(42)
for _ in range(50):
    study = random.randint(1, 10)
    sleep = random.randint(4, 10)
    grade = generate_grade(study, sleep)
    all_data.append(([study, sleep], grade))

# Step 2: Train (80% data)
split = int(len(all_data) * 0.8)
train_data = all_data[:split]
test_data = all_data[split:]

print(f"Training examples: {len(train_data)}")
print(f"Testing examples: {len(test_data)}")

# Step 3: Learn patterns (calculate threshold averages per grade)
grade_study_avg = {}
for features, grade in train_data:
    if grade not in grade_study_avg:
        grade_study_avg[grade] = []
    grade_study_avg[grade].append(features[0])

print()
print("Average study hours per grade (learned from training data):")
for grade in sorted(grade_study_avg.keys()):
    avg = sum(grade_study_avg[grade]) / len(grade_study_avg[grade])
    print(f"  Grade {grade}: {avg:.1f} hours average study time")

# Step 4: Simple predictor
def predict_grade(study_hours, sleep_hours):
    score = study_hours * 8 + sleep_hours * 4
    if score >= 80:
        return "A"
    elif score >= 60:
        return "B"
    elif score >= 40:
        return "C"
    else:
        return "D"

# Step 5: Evaluate
correct = sum(
    1 for features, actual in test_data
    if predict_grade(features[0], features[1]) == actual
)
accuracy = correct / len(test_data) * 100
print()
print(f"Model accuracy on test data: {accuracy:.1f}%")
print("Practice complete!")