README.md

requirements

WORK IN PROGRESS

1.) Python for Scientific Computing App

2.) Splunk ML Toolkit >=1.0

commands.conf

[corr]
filename = corr.py
chunked = true

corr.py

#!/usr/bin/env python
# Xander Johnson @metasyn

import sys
import exec_anaconda
try:
    exec_anaconda.exec_anaconda()
except Exception as e:
    import cexc
    cexc.abort(e)
    sys.exit(1)

# Import this to get our "chunked" handler
from cexc import BaseChunkHandler
import cexc

import pandas as pd

class Handler(BaseChunkHandler):
    def handler(self, metadata, data):
        """Get correlation matrix from a pandas dataframe"""

        if self.getinfo.get('preview', False):
            logger.debug('Not running in preview')
            return {'finished': True}

        def parse_args(raw_args):
            if len(raw_args) > 0:
                if 'method' not in raw_args[0]:
                    raise ValueError('The only accepted option is method: pearson, spearman, kendall. e.g. method=spearman')
                else:
                    try:
                        value = raw_args[0].split('=')[1]
                        if value == 'kendall' or value == 'spearman' or value == 'pearson':
                            return value
                        else:
                            raise ValueError('The only accepted values are: pearson, spearman, kendall. e.g. method=spearman')
                    except:
                        raise ValueError('The only accepted option is method: pearson, spearman, kendall. e.g. method=pearson')
            else:
                return 'pearson'

        if len(data) != 0:

            # get them args
            raw_args = self.getinfo['searchinfo']['raw_args']
            method = parse_args(raw_args)

            df = pd.DataFrame(data, dtype='float').corr(method=method)
            df.drop([col for col in df.columns if col.startswith('__')])
            df['(field)'] = df.columns

            output = df.to_dict('records')
            return({'finished': True}, output)

        if not metadata.get('finished', False):
            return {}
        else:
            return ({'finished': False}, data)

if __name__ == "__main__":
    Handler().run()