Fiddling with some introductory data mining

• Python 3.5+
• scikit-learn

./main.py --k=4 --limit=100 samples/income_tr.csv

• Generate a similarities table of the 4 closest records for the first 100 records in income_tr.csv using the default algorithm (composition of per-attribute or per-group similarities)

./main.py --k=4 --out results.csv --alt samples/income_tr.csv

• Generate a similarities table using the alternative algorithm (cosine similarity of a vectorization of each row) and write the resulting table to results.csv

./main.py --k=4 --knn --test samples/income_te.csv --out results.csv samples/income_tr.csv

• Run a k-Nearest Neighbor classifier against the test dataset (income_te) using the training dataset (income_tr) and outputting the results to results.csv. Uses majority votes for classification by default.

./main.py --k=4 --knn --weighted --test samples/income_te.csv --out results.csv samples/income_tr.csv

• Run a kNN classifier using weighted distances instead of majority votes against income_te and output the results to results.csv

Additional options available via --help