Pwml

Pwml stands for Python Wrappers for Machine Learning

Installation

• The Pwml package is published at https://pypi.org/project/Pwml/
• Install package: pip install Pwml

Samples

Hierarchical Model Training

The hierarchical model is basically defined like follow.

from Pwml.classifiers import hierarchical as hc
from Pwml.classifiers import features as fe

# (...)

model = hc.HierarchicalClassifierModel(
    model_name=model_name,
    experiment_name=experiment_name,
    input_features=[
        fe.InputFeature(feature_name='Style', feature_type='text'),
        fe.InputFeature(feature_name='Gender', feature_type='text'),
        fe.InputFeature(feature_name='Brand', feature_type='text'),
        fe.InputFeature(feature_name='Category', feature_type='text') ],
    output_feature_hierarchy=fe.OutputFeature(
        feature_name='Division',
        child_feature=fe.OutputFeature(feature_name='Class')))

It will fit n+1 models, where n is the number of distinct Division values:

• 1 model is fitted to handle the Division value.
• For each Division value a specific model is fitted to handle the possible Class values available under that Division.

Each model type is selected and fine-tuned separately.

Inference

This repository contains a sample solution implemented as a web-service using Flask, Flask-Restful, Flask-Cors and Flask-Wtf.

The web-service allows loading one or more pre-trained models and making a classification prediction based on a given sample along with its input features.