ID3 Decision Tree using Python

---

Assignment for COMP4388 (Machine Learning)

Main Functions

• calculate_entropy -> returns entropy for values inside feature column
• calculate_information_gain -> returns target feature entropy minus split feature entropy
• build_tree -> recursive id3 algorithm to build tree

---

Helper Functions

• prepare_data -> change from arff to pandas df
• get_unique_count -> pretty much np.unique
• get_max -> returns max element
• get_sum -> returns total sum
• predict -> create prediction column from tree built
• get_accuracy -> compares output of predict with target feature

---

Improvements: Cross Validation k = 3

I was getting ~76% accuracy originally, after adding cross validation I was able to improve it to ~95% accuracy.