Export a CSV file of GridSearchCV scores.

GridCV_scores_printer.py

import csv
import numpy as np
from sklearn.grid_search import GridSearchCV

def print_GridCV_scores(gs_clf, export_file):
	'''Exports a CSV of the GridCV scores.

	gs_clf: A GridSearchCV object which has been fitted
	export_file: A file path

	Example output (file content):
	mean,std,feat__words__ngram_range,feat__words__stop_words,clf__alpha
	0.56805074971164937,0.0082019735998974941,"(1, 2)",english,0.0001
	0.57189542483660127,0.0066384723824170488,"(1, 2)",None,0.0001
	0.56839677047289505,0.0082404203511470264,"(1, 3)",english,0.0001
	0.57306164295783668,0.0095988722286300399,"(1, 3)",None,0.0001
	0.53524285531205951,0.0026015635012174854,"(1, 2)",english,0.001
	0.53742150454953219,0.0031141868512110649,"(1, 2)",None,0.001
	0.53510829168268614,0.0032487504805843725,"(1, 3)",english,0.001
	0.53717160066641034,0.0034025374855825019,"(1, 3)",None,0.001
	'''
	with open(export_file, 'w') as outfile:
		csvwriter = csv.writer(outfile, delimiter=',')

		# Create the header using the parameter names 
		header = ["mean","std"]
		param_names = [param for param in gs_clf.param_grid]
		header.extend(param_names)

		csvwriter.writerow(header)

		for config in gs_clf.grid_scores_:
			# Get mean and standard deviation
			mean = config[1]
			std = np.std(config[2])
			row = [mean,std]

			# Get the list of parameter settings and add to row
			params = [str(p) for p in config[0].values()]
			row.extend(params)
			
			csvwriter.writerow(row)

grid_scores_ is deprecated, it is cv_results_ now.