Pattern Recognition of spam emails in a campus inbox with Decision Tree

Project info

• Field of studyInformatics and Information Systems
• LevelUndergraduate thesis (S1)
• Method typeClassification
• Estimated build time± 3 working days
• Toolsscikit-learn DecisionTreeClassifier, plot_tree, export_text
• PriceUS$19 - US$125

Research goal

Classify email class (spam or legitimate) from link count and message length with Decision Tree, then assess performance using accuracy and the confusion matrix.

Variables

Example result

• 0.689Accuracy
• 0.68Precision
• 0.739Recall
• 0.708F1-score
• 4Tree depth
• 10Leaves

The decision tree model classifies Email class (spam or legitima… (low/high) from Link count and Message length with test accuracy 0.689, F1 0.708 and AUC 0.788. Tree depth is 4 with 10 leaves.

Computed from illustrative synthetic data, not from a real study.

About Decision Tree

A decision tree recursively splits the data with if-then rules that are easy to read and visualise. It suits classification whose results must be explained as rules to non-technical readers.

Steps

1. 1Split into stratified training and test sets
2. 2Train DecisionTreeClassifier with a max_depth limit
3. 3Prune (ccp_alpha or min_samples_leaf) to reduce overfitting
4. 4Evaluate accuracy and F1 on the test set
5. 5Plot the tree and read its rules
6. 6Compute feature importance

Assumptions to check

• Deep trees overfit easily, so depth must be limited
• Results are sensitive to small changes in the data
• Classes should not be heavily imbalanced

Reading and reporting the results

Show the tree diagram and translate its main branches into meaningful rules, together with test accuracy. Mention that a single tree is less stable than ensembles.

Example Python code

A short starting script with this project's variable names. The delivered version is adapted to your data and fully commented.

import numpy as np
import matplotlib.pyplot as plt
from sklearn.tree import DecisionTreeClassifier, plot_tree, export_text
from sklearn.model_selection import train_test_split
from sklearn.metrics import accuracy_score, f1_score

rng = np.random.default_rng(21)
n = 300
link_count = rng.normal(0, 1, n)
message_length = rng.normal(0, 1, n)
email_class_spam_or_legitima = ((link_count > 0.2) & (message_length < 0.5)).astype(int)
flip = rng.random(n) < 0.05
email_class_spam_or_legitima[flip] = 1 - email_class_spam_or_legitima[flip]

X = np.column_stack([link_count, message_length])
X_tr, X_te, y_tr, y_te = train_test_split(X, email_class_spam_or_legitima, test_size=0.3, stratify=email_class_spam_or_legitima, random_state=21)
tree = DecisionTreeClassifier(max_depth=3, min_samples_leaf=5, random_state=21)
tree.fit(X_tr, y_tr)
pred = tree.predict(X_te)
print("Accuracy:", round(accuracy_score(y_te, pred), 3))
print("F1:", round(f1_score(y_te, pred), 3))
print("Importance:", tree.feature_importances_.round(3))
print(export_text(tree, feature_names=["x1", "x2"]))

plt.figure(figsize=(9, 5))
plot_tree(tree, feature_names=["x1", "x2"], class_names=["0", "1"], filled=True)
plt.show()

Packages and prices

Choose how much help you need. Each higher package includes everything in the lower ones. Prices exclude 11% VAT.

• Payment: 50% down payment before work starts, balance after the deliverables are handed over
• Tax: Prices exclude 11% VAT
• Data and academic integrity: Results are computed from your own research data; you write and stand behind your manuscript, we help with the analysis and your understanding of it
• Revisions: Up to 2 revisions are included in the price (Package 3: 3 rounds)

Frequently asked questions