Bayesbot is a naive bayesian classifier exposed as a simple web service. It is implemented as a Python/Flask web service using Redis as the back-end.

The web api is very simple and supports only one function, classification.

To classify a sample, send a get request to / in a format like this:

get /?f=feature1&f=feature2&f=feature3

Bayes bot will respond with a single text line containing the class that maps to the given features with the highest probability.

Training is accomplished with the bin/train script provided. It expects as input files containing one sample per line, with the class of each sample provided as the first field in each line. The first non-comment line of the file should be a comma separated list of the names of the features specified in the file. Sample training files are provided in the data directory.

Bayesbot is a Flask web service. To run bayesbot, first make sure redis is running on the target host and install redis and flask via easy_install/pip. Then start bayesbot directly as:

python bayesbot.py

Bayesbot currently expects redis to be running on localhost on the standard port but support for more flexible configuations will be implemented shortly.

To verify that your instance of bayesbot is running correctly, issue the following commands from the bayesbot root directory

bin/train data/mushroom.csv
bin/test data/mushroom.csv

You should see a series of string pairs. The left corresponds to the known correct class for the test sample and the right corresponds to the class predicted by bayesbot. At the end of the list of classes the test script will print the percentage accuracy of bayesbot, which should be > 0.95 for the mushroom sample.