Quickstart

This page shows a complete streaming loop with a model and a dataset stream.

1. Build a model

from aberrant.model.iforest import OnlineIsolationForest

model = OnlineIsolationForest(
    num_trees=50,
    max_leaf_samples=32,
    window_size=512,
)

2. Load a stream

from aberrant.stream.dataset import Dataset, load

dataset = load(Dataset.SHUTTLE)

3. Online train + score

from sklearn.metrics import average_precision_score

labels, scores = [], []

for i, (x, y) in enumerate(dataset.stream()):
    # Warm up on normal points only.
    if i < 2000:
        if y == 0:
            model.learn_one(x)
        continue

    model.learn_one(x)
    score = model.score_one(x)
    labels.append(y)
    scores.append(score)

print("PR-AUC:", average_precision_score(labels, scores))

4. Add thresholding

OnlineIsolationForest returns scores in [0, 1]. You can add a threshold model:

from aberrant.model import ThresholdModel

threshold = ThresholdModel(ceiling={"score": 0.8})

for x in stream_of_points:
    score = model.score_one(x)
    is_anomaly = threshold.score_one({"score": score}) == 1.0