Module 3: Sequential Pattern Mining

Code Along

Sequential Pattern Mining

Try to automatically find temporal patterns within the data set

Example: If student A watch the lecture videos, student A will read the comments later as well.

Generate the dataset

• Import
• Generate

Import the necessary package: random and PrefixSpan.

import random
from prefixspan import PrefixSpan

Randomly Generate the Dataset for SPM mining

# Possible activities
activities = ['GAMING', 'ON-TASK', 'OFF-TASK', 'BORED', 'FRUSTRATED']

# Function to generate random student activity sequences
def generate_student_data(num_students=20, max_sequence_length=6):
    student_data = []
    for _ in range(num_students):
        # Random sequence length between 3 and max_sequence_length
        sequence_length = random.randint(3, max_sequence_length)
        # Randomly select activities for the student (allowing repetition)
        sequence = random.choices(activities, k=sequence_length)  # Use random.choices instead of random.sample
        student_data.append(sequence)
    return student_data

# Generate random student data for 20 students
student_data = generate_student_data(20)
# Print the generated student data
print("Generated Student Data:")
for student in student_data:
    print(student)

Conduct the SPM

Conduct the analysis with PrefixSpan algorithm with the minimum support = 0.3

{python}
# Create a PrefixSpan object and run it on the student data
ps = PrefixSpan(student_data)

# Set a minimum support value (e.g., 0.3 means the pattern should appear in at least 30% of sequences)
min_support = 0.3
patterns = ps.frequent(min_support)

# Display the frequent sequential patterns
print("\nFrequent Sequential Patterns (with min_support=0.3):")
for pattern in patterns:
    print(pattern)