Implementation:Online ml River Datasets HTTP

Knowledge Sources	Online_ml_River
Domains	Online_Learning, Datasets, Binary_Classification, Anomaly_Detection
Last Updated	2026-02-08 16:00 GMT

Overview

Concrete dataset for binary classification provided by the River library.

Description

HTTP dataset of the KDD 1999 cup. The goal is to predict whether or not an HTTP connection is anomalous or not. The dataset only contains 2,211 (0.4%) positive labels, making it a highly imbalanced dataset suitable for anomaly detection tasks.

This dataset contains 567,498 samples with 3 features for binary classification tasks.

Usage

This dataset is useful for:

• Anomaly detection and network intrusion detection
• Handling imbalanced classification problems
• Evaluating classifiers on highly skewed data distributions
• Cybersecurity and network security applications

Code Reference

Source Location

• Repository: Online_ml_River
• File: river/datasets/http.py

Signature

class HTTP(base.RemoteDataset):
    def __init__(self):
        super().__init__(
            n_samples=567_498,
            n_features=3,
            task=base.BINARY_CLF,
            url="https://maxhalford.github.io/files/datasets/kdd99_http.zip",
            size=32_400_738,
            filename="kdd99_http.csv",
        )

    def _iter(self):
        converters = {
            "duration": float,
            "src_bytes": float,
            "dst_bytes": float,
            "service": int,
        }
        return stream.iter_csv(self.path, target="service", converters=converters)

Import

from river import datasets
dataset = datasets.HTTP()

I/O Contract

Inputs

Name	Type	Required	Description
(none)	—	—	No parameters needed

Outputs

Name	Type	Description
iter()	tuple(dict, int)	Yields (features_dict, target) pairs where target indicates anomalous connections

Dataset Properties

Property	Value
Number of samples	567,498
Number of features	3
Task	Binary classification (anomaly detection)
Format	CSV (compressed)
Size	32,400,738 bytes
Class imbalance	Only 0.4% positive labels (2,211 anomalies)

Features

The dataset includes the following features:

• duration: Duration of the connection (float)
• src_bytes: Number of bytes from source to destination (float)
• dst_bytes: Number of bytes from destination to source (float)
• service: Target variable indicating if connection is anomalous (integer)

Usage Examples

from river import datasets

dataset = datasets.HTTP()
for x, y in dataset:
    print(x, y)
    break

References

• HTTP (KDDCUP99) dataset

Related Pages

• Environment:Online_ml_River_Python_Runtime_Environment