jubakit 0.1.0

Jubatus Toolkit

jubakit: Jubatus Toolkit

jubakit is a Python module to access Jubatus features easily. jubakit can be used in conjunction with scikit-learn so that you can use powerful features like cross validation and model evaluation.

Install

pip install jubakit

Requirements

• Python 2.6, 2.7, 3.3, 3.4 or 3.5.

• Jubatus needs to be installed.

• Although not mandatory, installing scikit-learn is required to use some features like K-fold cross validation.

Quick Start

The following example shows how to perform train/classify using CSV dataset.

from jubakit.classifier import Classifier, Schema, Dataset, Config
from jubakit.loader.csv import CSVLoader

# Load a CSV file.
loader = CSVLoader('iris.csv')

# Define types for each column in the CSV file.
schema = Schema({
  'Species': Schema.LABEL,
}, Schema.NUMBER)

# Get the shuffled dataset.
dataset = Dataset(loader, schema).shuffle()

# Run the classifier service (`jubaclassifier` process).
classifier = Classifier.run(Config())

# Train the classifier.
for _ in classifier.train(dataset): pass

# Classify using the trained classifier.
for (idx, label, result) in classifier.classify(dataset):
  print("true label: {0}, estimated label: {1}".format(label, result[0][0]))

Examples by Topics

See the example directory for working examples.

Example	Topics	Requires scikit-learn
classifier_csv.py	Handling CSV file and numeric features
classifier_shogun.py	Handling CSV file and string features
classifier_digits.py	Handling toy dataset (digits)	✓
classifier_libsvm.py	Handling LIBSVM file	✓
classifier_kfold.py	K-fold cross validation and metrics	✓
classifier_parameter.py	Finding best hyper parameter	✓
classifier_bulk.py	Bulk Train-Test Classifier
classifier_twitter.py	Handling Twitter Streams

Concepts

• Loader fetches data from various data sources (e.g., CSV file, RDBMS, MQ, Twitter stream, etc.) in key-value format.

• Schema defines the data type (string feature, numeric feature, ground truth (label), etc.) for each keys of data loaded by Loader.

• Dataset is an abstract representation of a sequence of data that binds Loader and Schema.

• Service receives Dataset and make update/analyze RPC call to Jubatus servers.

License

MIT License