Select Git revision
ExecClassifMonoView.py 14.95 KiB
#!/usr/bin/env python
""" Execution: Script to perform a MonoView classification """
# Import built-in modules
import argparse # for command line arguments
import datetime # for TimeStamp in CSVFile
import os # to geth path of the running script
import time # for time calculations
import operator
# Import 3rd party modules
import numpy as np # for reading CSV-files and Series
import logging # To create Log-Files
from sklearn import metrics # For stastics on classification
import h5py
# Import own modules
import MonoviewUtils # Functions for classification
import ExportResults # Functions to render results
import MonoviewClassifiers
import Metrics
from analyzeResult import execute
from utils.Dataset import getV, getValue, extractSubset
# Author-Info
__author__ = "Nikolas Huelsmann, Baptiste BAUVIN"
__status__ = "Prototype" # Production, Development, Prototype
__date__ = 2016-03-25
def ExecMonoview_multicore(directory, name, labelsNames, classificationIndices, KFolds, datasetFileIndex, databaseType, path, statsIter, randomState, hyperParamSearch="randomizedSearch",
metrics=[["accuracy_score", None]], nIter=30, **args):
DATASET = h5py.File(path+name+str(datasetFileIndex)+".hdf5", "r")
kwargs = args["args"]
views = [DATASET.get("View"+str(viewIndex)).attrs["name"] for viewIndex in range(DATASET.get("Metadata").attrs["nbView"])]
neededViewIndex = views.index(kwargs["feat"])
X = DATASET.get("View"+str(neededViewIndex))
Y = DATASET.get("Labels").value
return ExecMonoview(directory, X, Y, name, labelsNames, classificationIndices, KFolds, 1, databaseType, path, statsIter, randomState, hyperParamSearch=hyperParamSearch,
metrics=metrics, nIter=nIter, **args)
def ExecMonoview(directory, X, Y, name, labelsNames, classificationIndices, KFolds, nbCores, databaseType, path, statsIter, randomState, hyperParamSearch="randomizedSearch",
metrics=[["accuracy_score", None]], nIter=30, **args):
logging.debug("Start:\t Loading data")
try:
kwargs = args["args"]
except:
kwargs = args
t_start = time.time()
feat = X.attrs["name"]
CL_type = kwargs["CL_type"]
nbClass = kwargs["nbClass"]
X = getValue(X)
datasetLength = X.shape[0]
logging.debug("Done:\t Loading data")
# Determine the Database to extract features
print KFolds
logging.debug("Info:\t Classification - Database:" + str(name) + " Feature:" + str(feat) + " train_size:"
+ str(len(classificationIndices[0])) + ", CrossValidation k-folds: " + str(KFolds.n_splits) + ", cores:"
+ str(nbCores) + ", algorithm : " + CL_type)
# y_trains = []
# y_tests = []
# y_train_preds = []
# y_test_preds = []
trainIndices, testIndices = classificationIndices
# for iterationStat in range(statsIter):
# Calculate Train/Test data
logging.debug("Start:\t Determine Train/Test split") #+" for iteration "+str(iterationStat+1)
# testIndices = MonoviewUtils.splitDataset(Y, nbClass, classificationIndices, datasetLength, randomState)
# trainIndices = [i for i in range(datasetLength) if i not in testIndices]
X_train = extractSubset(X, trainIndices)
X_test = extractSubset(X, testIndices)
y_train = Y[trainIndices]
y_test = Y[testIndices]
logging.debug("Info:\t Shape X_train:" + str(X_train.shape) + ", Length of y_train:" + str(len(y_train)))
logging.debug("Info:\t Shape X_test:" + str(X_test.shape) + ", Length of y_test:" + str(len(y_test)))
logging.debug("Done:\t Determine Train/Test split")
classifierModule = getattr(MonoviewClassifiers, CL_type)
if hyperParamSearch != "None":
classifierHPSearch = getattr(classifierModule, hyperParamSearch)
logging.debug("Start:\t RandomSearch best settings with "+str(nIter)+" iterations for "+CL_type)
cl_desc = classifierHPSearch(X_train, y_train, randomState, KFolds=KFolds, nbCores=nbCores,
metric=metrics[0], nIter=nIter)
clKWARGS = dict((str(index), desc) for index, desc in enumerate(cl_desc))
logging.debug("Done:\t RandomSearch best settings")
else:
clKWARGS = kwargs[kwargs["CL_type"]+"KWARGS"]
logging.debug("Start:\t Training")
cl_res = classifierModule.fit(X_train, y_train, randomState, NB_CORES=nbCores, **clKWARGS)
logging.debug("Done:\t Training")
logging.debug("Start:\t Predicting")
# Stats Result
y_train_pred = cl_res.predict(X_train)
y_test_pred = cl_res.predict(X_test)
# y_trains.append(y_train)
# y_train_preds.append(y_train_pred)
# y_tests.append(y_test)
# y_test_preds.append(y_test_pred)
full_labels = cl_res.predict(X)
logging.debug("Done:\t Predicting")
t_end = time.time() - t_start
logging.debug("Info:\t Time for training and predicting: " + str(t_end) + "[s]")
logging.debug("Start:\t Getting Results")
stringAnalysis, imagesAnalysis, metricsScores = execute(name, classificationIndices, KFolds, nbCores, hyperParamSearch, metrics, nIter, feat, CL_type,
clKWARGS, labelsNames, X.shape,
y_train, y_train_pred, y_test, y_test_pred, t_end, statsIter, randomState)
cl_desc = [value for key, value in sorted(clKWARGS.iteritems())]
logging.debug("Done:\t Getting Results")
logging.info(stringAnalysis)
labelsString = "-".join(labelsNames)
timestr = time.strftime("%Y%m%d-%H%M%S")
CL_type_string = CL_type
outputFileName = directory + timestr + "Results-" + CL_type_string + "-" + labelsString + \
'-learnRate' + str(len(classificationIndices)) + '-' + name + "-" + feat
outputTextFile = open(outputFileName + '.txt', 'w')
outputTextFile.write(stringAnalysis)
outputTextFile.close()
if imagesAnalysis is not None:
for imageName in imagesAnalysis:
if os.path.isfile(outputFileName + imageName + ".png"):
for i in range(1,20):
testFileName = outputFileName + imageName + "-" + str(i) + ".png"
if os.path.isfile(testFileName ) != True:
imagesAnalysis[imageName].savefig(testFileName)
break
imagesAnalysis[imageName].savefig(outputFileName + imageName + '.png')
logging.info("Done:\t Result Analysis")
viewIndex = args["viewIndex"]
return viewIndex, [CL_type, cl_desc+[feat], metricsScores, full_labels]
# # Classification Report with Precision, Recall, F1 , Support
# logging.debug("Info:\t Classification report:")
# filename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + name + "-" + feat + "-Report"
# logging.debug("\n" + str(metrics.classification_report(y_test, y_test_pred, labels = range(0,len(classLabelsDesc.name)), target_names=classLabelsNamesList)))
# scores_df = ExportResults.classification_report_df(directory, filename, y_test, y_test_pred, range(0, len(classLabelsDesc.name)), classLabelsNamesList)
#
# # Create some useful statistcs
# logging.debug("Info:\t Statistics:")
# filename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + name + "-" + feat + "-Stats"
# stats_df = ExportResults.classification_stats(directory, filename, scores_df, accuracy_score)
# logging.debug("\n" + stats_df.to_string())
#
# # Confusion Matrix
# logging.debug("Info:\t Calculate Confusionmatrix")
# filename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + name + "-" + feat + "-ConfMatrix"
# df_conf_norm = ExportResults.confusion_matrix_df(directory, filename, y_test, y_test_pred, classLabelsNamesList)
# filename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + name + "-" + feat + "-ConfMatrixImg"
# ExportResults.plot_confusion_matrix(directory, filename, df_conf_norm)
#
# logging.debug("Done:\t Statistic Results")
#
#
# # Plot Result
# logging.debug("Start:\t Plot Result")
# np_score = ExportResults.calcScorePerClass(y_test, cl_res.predict(X_test).astype(int))
# ### directory and filename the same as CSV Export
# filename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + name + "-" + feat + "-Score"
# ExportResults.showResults(directory, filename, name, feat, np_score)
# logging.debug("Done:\t Plot Result")
# return [CL_type, accuracy_score, cl_desc]
if __name__=='__main__':
parser = argparse.ArgumentParser(
description='This methods permits to execute a multiclass classification with one single view. At this point the used classifier is a RandomForest. The GridSearch permits to vary the number of trees and CrossValidation with k-folds. The result will be a plot of the score per class and a CSV with the best classifier found by the GridSearch.',
formatter_class=argparse.ArgumentDefaultsHelpFormatter)
groupStandard = parser.add_argument_group('Standard arguments')
groupStandard.add_argument('-log', action='store_true', help='Use option to activate Logging to Console')
groupStandard.add_argument('--type', metavar='STRING', action='store', help='Type of Dataset', default=".hdf5")
groupStandard.add_argument('--name', metavar='STRING', action='store', help='Name of Database (default: %(default)s)', default='DB')
groupStandard.add_argument('--feat', metavar='STRING', action='store', help='Name of Feature for Classification (default: %(default)s)', default='RGB')
groupStandard.add_argument('--pathF', metavar='STRING', action='store', help='Path to the views (default: %(default)s)', default='Results-FeatExtr/')
groupStandard.add_argument('--fileCL', metavar='STRING', action='store', help='Name of classLabels CSV-file (default: %(default)s)', default='classLabels.csv')
groupStandard.add_argument('--fileCLD', metavar='STRING', action='store', help='Name of classLabels-Description CSV-file (default: %(default)s)', default='classLabels-Description.csv')
groupStandard.add_argument('--fileFeat', metavar='STRING', action='store', help='Name of feature CSV-file (default: %(default)s)', default='feature.csv')
groupClass = parser.add_argument_group('Classification arguments')
groupClass.add_argument('--CL_type', metavar='STRING', action='store', help='Classifier to use', default="RandomForest")
groupClass.add_argument('--CL_CV', metavar='INT', action='store', help='Number of k-folds for CV', type=int, default=10)
groupClass.add_argument('--CL_Cores', metavar='INT', action='store', help='Number of cores, -1 for all', type=int, default=1)
groupClass.add_argument('--CL_split', metavar='FLOAT', action='store', help='Split ratio for train and test', type=float, default=0.9)
groupClass.add_argument('--CL_metrics', metavar='STRING', action='store',
help='Determine which metrics to use, separate with ":" if multiple, if empty, considering all', default='')
groupClassifier = parser.add_argument_group('Classifier Config')
groupClassifier.add_argument('--CL_config', metavar='STRING', nargs="+", action='store', help='GridSearch: Determine the trees', default=['25:75:125:175'])
# groupSVMLinear = parser.add_argument_group('SVC arguments')
# groupSVMLinear.add_argument('--CL_SVML_C', metavar='STRING', action='store', help='GridSearch : Penalty parameters used', default='1:10:100:1000')
#
# groupSVMRBF = parser.add_argument_group('SVC arguments')
# groupSVMRBF.add_argument('--CL_SVMR_C', metavar='STRING', action='store', help='GridSearch : Penalty parameters used', default='1:10:100:1000')
#
# groupRF = parser.add_argument_group('Decision Trees arguments')
# groupRF.add_argument('--CL_DT_depth', metavar='STRING', action='store', help='GridSearch: Determine max depth for Decision Trees', default='1:3:5:7')
#
# groupSGD = parser.add_argument_group('SGD')
# groupSGD.add_argument('--CL_SGD_alpha', metavar='STRING', action='store', help='GridSearch: Determine alpha for SGDClassifier', default='0.1:0.2:0.5:0.9')
# groupSGD.add_argument('--CL_SGD_loss', metavar='STRING', action='store', help='GridSearch: Determine loss for SGDClassifier', default='log')
# groupSGD.add_argument('--CL_SGD_penalty', metavar='STRING', action='store', help='GridSearch: Determine penalty for SGDClassifier', default='l2')
args = parser.parse_args()
# RandomForestKWARGS = {"classifier__n_estimators":map(int, args.CL_RF_trees.split())}
# SVMLinearKWARGS = {"classifier__C":map(int,args.CL_SVML_C.split(":"))}
# SVMRBFKWARGS = {"classifier__C":map(int,args.CL_SVMR_C.split(":"))}
# DecisionTreeKWARGS = {"classifier__max_depth":map(int,args.CL_DT_depth.split(":"))}
# SGDKWARGS = {"classifier__alpha" : map(float,args.CL_SGD_alpha.split(":")), "classifier__loss":args.CL_SGD_loss.split(":"),
# "classifier__penalty":args.CL_SGD_penalty.split(":")}
classifierKWARGS = dict((key, value) for key, value in enumerate([arg.split(":") for arg in args.CL_config]))
### Main Programm
# Configure Logger
directory = os.path.dirname(os.path.abspath(__file__)) + "/Results-ClassMonoView/"
logfilename = datetime.datetime.now().strftime("%Y_%m_%d") + "-CMV-" + args.name + "-" + args.feat + "-LOG"
logfile = directory + logfilename
if os.path.isfile(logfile + ".log"):
for i in range(1,20):
testFileName = logfilename + "-" + str(i) + ".log"
if os.path.isfile(directory + testFileName)!=True:
logfile = directory + testFileName
break
else:
logfile = logfile + ".log"
logging.basicConfig(format='%(asctime)s %(levelname)s: %(message)s', filename=logfile, level=logging.DEBUG, filemode='w')
if(args.log):
logging.getLogger().addHandler(logging.StreamHandler())
# Read the features
logging.debug("Start:\t Read " + args.type + " Files")
if args.type == ".csv":
X = np.genfromtxt(args.pathF + args.fileFeat, delimiter=';')
Y = np.genfromtxt(args.pathF + args.fileCL, delimiter=';')
elif args.type == ".hdf5":
dataset = h5py.File(args.pathF + args.name + ".hdf5", "r")
viewsDict = dict((dataset.get("View"+str(viewIndex)).attrs["name"], viewIndex) for viewIndex in range(dataset.get("Metadata").attrs["nbView"]))
X = dataset["View"+str(viewsDict[args.feat])][...]
Y = dataset["Labels"][...]
logging.debug("Info:\t Shape of Feature:" + str(X.shape) + ", Length of classLabels vector:" + str(Y.shape))
logging.debug("Done:\t Read CSV Files")
arguments = {args.CL_type+"KWARGS": classifierKWARGS, "feat":args.feat,"fileFeat": args.fileFeat,
"fileCL": args.fileCL, "fileCLD": args.fileCLD, "CL_type": args.CL_type}
ExecMonoview(X, Y, args.name, args.CL_split, args.CL_CV, args.CL_Cores, args.type, args.pathF,
metrics=args.CL_metrics, **arguments)