Popsom7

Overview

A fast, user-friendly implementation of self-organizing maps (SOMs) with a number of distinguishing features:

1. Support for both Python and R.

2. Easy to use interfaces for building and evaluating self-organizing maps:

   • An interface that works the same on both the R and the Python platforms
   • An interface that is sklearn compatible, allowing you to leverage the power and convenience of the sklearn framework in Python.

3. Automatic centroid detection and visualization using starbursts.

4. Two models of the data: (a) a self organizing map model, (b) a centroid based clustering model.

5. A number of easily accessible quality metrics.

6. An implementation of the training algorithm based on tensor algebra.

Installation

You can install popsom7 via pip:

pip install popsom7

Usage

Below is a quick example using the popsom sklearnapi interface.

from popsom7.sklearnapi import SOM
import pandas as pd
from sklearn import datasets

iris = datasets.load_iris()
X = pd.DataFrame(iris.data, columns=iris.feature_names)
y = pd.DataFrame(iris.target_names[iris.target],columns=['species'])

# Create and fit the SOM model
som = SOM(xdim=20, ydim=15, train=100000, seed=42).fit(X, y)

# View a summary of the SOM
som.summary()

# Display the starburst (heat map) representation
som.starburst()

Here is the same example written in the API based on the R API.

from popsom7.maputils import map_build, map_summary, map_starburst
import pandas as pd
from sklearn import datasets   

iris = datasets.load_iris()
X = pd.DataFrame(iris.data, columns=iris.feature_names)
y = pd.DataFrame(iris.target_names[iris.target],columns=['species'])

# Build the map
som_map = map_build(X, labels=y, xdim=20, ydim=15, train=100000, seed=42)

# View a summary of the SOM
map_summary(som_map)

# Display the starburst (heat map) representation
map_starburst(som_map)

For more details please see the project homepage