exstracs_output.py

"""
Name:        ExSTraCS_OutputFileManager.py
Authors:     Ryan Urbanowicz - Written at Dartmouth College, Hanover, NH, USA
Contact:     ryan.j.urbanowicz@darmouth.edu
Created:     April 25, 2014
Modified:    August 25,2014
Description: This module contains the methods for generating the different output files generated by ExSTraCS.
             These files are generated at each learning checkpoint, and the last iteration.  These include...
             writePopStats:   Summary of the population statistics
             writePop:        Outputs a snapshot of the entire rule population including classifier conditions, classes, and parameters.
             attCo_Occurence: Calculates and outputs co-occurrence scores for each attribute pair in the dataset.

---------------------------------------------------------------------------------------------------------------------------------------------------------
ExSTraCS V2.0: Extended Supervised Tracking and Classifying System - An advanced LCS designed specifically for complex, noisy classification/data mining tasks,
such as biomedical/bioinformatics/epidemiological problem domains.  This algorithm should be well suited to any supervised learning problem involving
classification, prediction, data mining, and knowledge discovery.  This algorithm would NOT be suited to function approximation, behavioral modeling,
or other multi-step problems.  This LCS algorithm is most closely based on the "UCS" algorithm, an LCS introduced by Ester Bernado-Mansilla and
Josep Garrell-Guiu (2003) which in turn is based heavily on "XCS", an LCS introduced by Stewart Wilson (1995).

Copyright (C) 2014 Ryan Urbanowicz
This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the
Free Software Foundation; either version 3 of the License, or (at your option) any later version.

This program is distributed in the hope that it will be useful but WITHOUT ANY WARRANTY; without even the implied warranty of MERCHANTABLILITY
or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License for more details.

You should have received a copy of the GNU General Public License along with this program; if not, write to the Free Software Foundation,
Inc., 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301 USA
---------------------------------------------------------------------------------------------------------------------------------------------------------
"""

#Import Required Modules-------------------------------
from exstracs_constants import *
from exstracs_at import *
import copy
#------------------------------------------------------

class OutputFileManager:

    def writePopStats(self, outFile, trainEval, testEval, exploreIter, pop, correct):
        """ Makes output text file which includes all of the parameter settings used in the run as well as all of the evaluation stats including Time Track Output. """
        if cons.outputSummary:
            try:
                popStatsOut = open(outFile + '_'+ str(exploreIter)+'_PopStats.txt','w') # Outputs Population run stats
                print("Writing Population Statistical Summary File...")
            except IOError as xxx_todo_changeme:
                (errno, strerror) = xxx_todo_changeme.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise

            #Evaluation of pop
            popStatsOut.write("Performance Statistics:------------------------------------------------------------------------\n")
            popStatsOut.write("Training Accuracy\tTesting Accuracy\tTraining Coverage\tTesting Coverage\n")

            if cons.testFile != 'None':
                popStatsOut.write(str(trainEval[0])+"\t")
                popStatsOut.write(str(testEval[0])+"\t")
                popStatsOut.write(str(trainEval[1]) +"\t")
                popStatsOut.write(str(testEval[1])+"\n\n")
            elif cons.trainFile != 'None':
                popStatsOut.write(str(trainEval[0])+"\t")
                popStatsOut.write("NA\t")
                popStatsOut.write(str(trainEval[1]) +"\t")
                popStatsOut.write("NA\n\n")
            else:
                popStatsOut.write("NA\t")
                popStatsOut.write("NA\t")
                popStatsOut.write("NA\t")
                popStatsOut.write("NA\n\n")

            popStatsOut.write("Population Characterization:------------------------------------------------------------------------\n")
            popStatsOut.write("MacroPopSize\tMicroPopSize\tGenerality\n")
            popStatsOut.write(str(len(pop.popSet))+"\t"+ str(pop.microPopSize)+"\t"+str(pop.aveGenerality)+"\n\n")

            popStatsOut.write("SpecificitySum:------------------------------------------------------------------------\n")
            headList = cons.env.formatData.trainHeaderList #preserve order of original dataset

            for i in range(len(headList)):
                if i < len(headList)-1:
                    popStatsOut.write(str(headList[i])+"\t")
                else:
                    popStatsOut.write(str(headList[i])+"\n")

            # Prints out the Specification Sum for each attribute
            for i in range(len(pop.attributeSpecList)):
                if i < len(pop.attributeSpecList)-1:
                    popStatsOut.write(str(pop.attributeSpecList[i])+"\t")
                else:
                    popStatsOut.write(str(pop.attributeSpecList[i])+"\n")

            popStatsOut.write("\nAccuracySum:------------------------------------------------------------------------\n")
            for i in range(len(headList)):
                if i < len(headList)-1:
                    popStatsOut.write(str(headList[i])+"\t")
                else:
                    popStatsOut.write(str(headList[i])+"\n")

            # Prints out the Accuracy Weighted Specification Count for each attribute
            for i in range(len(pop.attributeAccList)):
                if i < len(pop.attributeAccList)-1:
                    popStatsOut.write(str(pop.attributeAccList[i])+"\t")
                else:
                    popStatsOut.write(str(pop.attributeAccList[i])+"\n")

            if cons.onlyRC: # When RC ONLY, there is no AttributeTrackingGlobalSums
                popStatsOut.write("\nAttributeTrackingGlobalSums:----Rule Compaction ONLY, Attribute Tracking not loaded-----------------\n")
                for i in range(len(headList)):
                    if i < len(headList)-1:
                        popStatsOut.write(str(headList[i])+"\t")
                    else:
                        popStatsOut.write(str(headList[i])+"\n")
                for i in range(len(headList)):
                    if i < len(headList)-1:
                        popStatsOut.write(str(0.0)+"\t")
                    else:
                        popStatsOut.write(str(0.0)+"\n")
            elif cons.doAttributeTracking:
                popStatsOut.write("\nAttributeTrackingGlobalSums:------------------------------------------------------------------------\n")
                for i in range(len(headList)):
                    if i < len(headList)-1:
                        popStatsOut.write(str(headList[i])+"\t")
                    else:
                        popStatsOut.write(str(headList[i])+"\n")

                sumGlobalAttTrack = cons.AT.sumGlobalAttTrack()
                for i in range(len(sumGlobalAttTrack)):
                    if i < len(sumGlobalAttTrack)-1:
                        popStatsOut.write(str(sumGlobalAttTrack[i])+"\t")
                    else:
                        popStatsOut.write(str(sumGlobalAttTrack[i])+"\n")
            else:
                popStatsOut.write("\nAttributeTrackingGlobalSums:----Tracking not applied!-----------------------------------------------\n")
                for i in range(len(headList)):
                    if i < len(headList)-1:
                        popStatsOut.write(str(headList[i])+"\t")
                    else:
                        popStatsOut.write(str(headList[i])+"\n")
                for i in range(len(headList)):
                    if i < len(headList)-1:
                        popStatsOut.write(str(0.0)+"\t")
                    else:
                        popStatsOut.write(str(0.0)+"\n")

            #Time Track ---------------------------------------------------------------------------------------------------------
            popStatsOut.write("\nRun Time(in minutes):------------------------------------------------------------------------\n")
            popStatsOut.write(cons.timer.reportTimes())
            popStatsOut.write("\nCorrectTrackerSave:------------------------------------------------------------------------\n")
            for i in range(len(correct)):
                popStatsOut.write(str(correct[i])+"\t")
#             popStatsOut.write("\nMax/MinStateFrequencies:------------------------------------------------------------------------\n")
#             for i in range(len(cons.env.formatData.maxFreq)):
#                 popStatsOut.write(str(cons.env.formatData.maxFreq[i])+"\t")
#             popStatsOut.write('\n')
#             for i in range(len(cons.env.formatData.maxFreq)):
#                 popStatsOut.write(str(cons.env.formatData.minFreq[i])+"\t")
#             popStatsOut.write("\nBestCoverDiff:------------------------------------------------------------------------\n")
# #             for each in cons.env.formatData.bestCoverDiff:
# #                 popStatsOut.write(str(each)+"\t"+str(cons.env.formatData.bestCoverDiff[each][0])+"\t"+str(cons.env.formatData.bestCoverDiff[each][1])+"\t"+str(cons.env.formatData.bestCoverDiff[each][2])+"\t"+str(cons.env.formatData.bestCoverDiff[each][3])+"\n")
#             popStatsOut.write(str(cons.env.formatData.bestCoverDiff[0])+"\t"+str(cons.env.formatData.bestCoverDiff[1])+"\t"+str(cons.env.formatData.bestCoverDiff[2])+"\t"+str(cons.env.formatData.bestCoverDiff[3])+"\n")
            popStatsOut.write("\nGlobalEpochStatus:------------------------------------------------------------------------\n")
            popStatsOut.write(str(cons.firstEpochComplete))
            #popStatsOut.write(str(cons.firstEpochComplete)+"\t"+str(cons.epochPoolFull))
            popStatsOut.write("\nParetoFronts:------------------------------------------------------------------------\n")
#             popStatsOut.write("Incomplete Epoch:\n")
#             for each in cons.env.formatData.phenotypeList:
#                 popStatsOut.write("Class: "+ str(each)+ "\n")
#                 popStatsOut.write(self.tabPareto(cons.env.formatData.bestCoverDiff[each][4].paretoFrontAcc) + "\n")
#                 popStatsOut.write(self.tabPareto(cons.env.formatData.bestCoverDiff[each][4].paretoFrontRawCov) + "\n")
#             popStatsOut.write("Complete Epoch:\n")
#             for each in cons.env.formatData.phenotypeList:
#                 popStatsOut.write("Class: "+ str(each)+ "\n")
#                 popStatsOut.write(self.tabPareto(cons.env.formatData.bestCoverDiff[each][5].paretoFrontAcc) + "\n")
#                 popStatsOut.write(self.tabPareto(cons.env.formatData.bestCoverDiff[each][5].paretoFrontRawCov) + "\n")

            popStatsOut.write("Incomplete Epoch:\n")
            popStatsOut.write(self.tabPareto(cons.env.formatData.necFront.paretoFrontAcc) + "\n")
            popStatsOut.write(self.tabPareto(cons.env.formatData.necFront.paretoFrontRawCov) + "\n")

            popStatsOut.write("Complete Epoch:\n")
            popStatsOut.write(self.tabPareto(cons.env.formatData.ecFront.paretoFrontAcc) + "\n")
            popStatsOut.write(self.tabPareto(cons.env.formatData.ecFront.paretoFrontRawCov) + "\n")
            try:
                popStatsOut.close()
            except IOError as xxx_todo_changeme1:
                (errno, strerror) = xxx_todo_changeme1.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise
        else:
            pass


    def tabPareto(self, theList):
        theString = ''
        for each in theList:
            theString += str(each)+'\t'
        return theString


    def writePop(self, outFile, exploreIter, pop):
        """ Writes a tab delimited text file specifying the evolved rule population, including conditions, phenotypes, and all rule parameters. """
        if cons.outputPopulation:
            try:
                rulePopOut = open(outFile + '_'+ str(exploreIter)+'_RulePop.txt','w') # Outputs tab delimited text file of rule population and respective rule stats
                print("Writing Population as Data File...")
            except IOError as xxx_todo_changeme2:
                (errno, strerror) = xxx_todo_changeme2.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise
            rulePopOut.write("Specified\tCondition\tPhenotype\tFitness\tAccuracy\tNumerosity\tAveMatchSetSize\tTimeStampGA\tInitTimeStamp\tSpecificity\tDeletionProb\tCorrectCount\tMatchCount\tCorrectCover\tMatchCover\tEpochComplete\tAccuracyComp\tCoverDiff\tIndFitness\tFitness\tGlobalFitness\tTotalFreq\n")

            #Write each classifier--------------------------------------------------------------------------------------------------------------------------------------
            for cl in pop.popSet:
                rulePopOut.write(str(cl.printClassifier()))
            try:
                rulePopOut.close()
            except IOError as xxx_todo_changeme3:
                (errno, strerror) = xxx_todo_changeme3.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise
        else:
            pass


#    def writePop(self, outFile, exploreIter, pop):
#        """ Writes a tab delimited text file specifying the evolved rule population, including conditions, phenotypes, and all rule parameters. """
#        if cons.outputPopulation:
#            try:
#                rulePopOut = open(outFile + '_'+ str(exploreIter)+'_RulePop.txt','w') # Outputs tab delimited text file of rule population and respective rule stats
#                print "Writing Population as Data File..."
#            except IOError, (errno, strerror):
#                print ("I/O error(%s): %s" % (errno, strerror))
#                raise
#            rulePopOut.write("Specified\tCondition\tPhenotype\tFitness\tAccuracy\tNumerosity\tAveMatchSetSize\tTimeStampGA\tInitTimeStamp\tSpecificity\tDeletionProb\tCorrectCount\tMatchCount\tCorrectCover\tMatchCover\tEpochComplete\n")
#
#            #Write each classifier--------------------------------------------------------------------------------------------------------------------------------------
#            for cl in pop.popSet:
#                rulePopOut.write(str(cl.printClassifier()))
#            try:
#                rulePopOut.close()
#            except IOError, (errno, strerror):
#                print ("I/O error(%s): %s" % (errno, strerror))
#                raise
#        else:
#            pass


    def attCo_Occurence(self, outFile, exploreIter, pop):
        """ Calculates pairwise attribute co-occurence througout all rules in the population."""
        if cons.outputAttCoOccur:
            print("Calculating Attribute Co-occurence Scores...")
            dataLink = cons.env.formatData
            dim = dataLink.numAttributes
            maxAtts = 50  #Test 10
            attList = []
            #-------------------------------------------------------
            # IDENTIFY ATTRIBUBTES FOR CO-OCCRRENCE EVALUATION
            #-------------------------------------------------------
            if dim <= maxAtts:
                for i in range(0,dim):
                    attList.append(i)
            else:
                tempList = copy.deepcopy(pop.attributeSpecList)
                tempList = sorted(tempList,reverse=True)
                maxVal = tempList[maxAtts]
                overflow = []
                for i in range(0,dim):
                    if pop.attributeSpecList[i] >= maxVal: #get roughly the top 50 specified attributes. (may grab some extras, depending on
                        attList.append(i)
                        if pop.attributeSpecList[i] == maxVal:
                            overflow.append(i)
                while len(attList) > maxAtts:
                    target = random.choice(overflow)
                    attList.remove(target)
                    overflow.remove(target)
                    print(attList)

            #-------------------------------------------------------
            # CO-OCCRRENCE EVALUATION
            #-------------------------------------------------------
            comboList = []
            castList = [None,None,0,0] #att1, att2, Specificity, Accuracy Weighted Specificity
            count = 0
            dim = dataLink.numAttributes
            #Specify all attribute pairs.
            for i in range(0, len(attList)-1):
                for j in range(i+1,len(attList)):
                    comboList.append(copy.deepcopy(castList))
                    comboList[count][0] = dataLink.trainHeaderList[attList[i]]
                    comboList[count][1] = dataLink.trainHeaderList[attList[j]]
                    count += 1

            for cl in pop.popSet:
                count = 0
                if not cl.isTree:
                    for i in range(len(attList)-1):
                        for j in range(i+1,len(attList)):
                            if attList[i] in cl.specifiedAttList and attList[j] in cl.specifiedAttList:
                                comboList[count][2] += cl.numerosity
                                comboList[count][3] += cl.numerosity * cl.accuracy
                            count += 1

            tupleList = []
            for i in comboList:
                tupleList.append((i[0],i[1],i[2],i[3]))
            sortedComboList = sorted(tupleList,key=lambda test: test[3], reverse=True)
            print("Writing Attribute Co-occurence scores as data file...")
            f = open(outFile + '_'+ str(exploreIter)+ '_CO.txt', 'w')
            for i in range(len(sortedComboList)):
                for j in range(len(sortedComboList[0])): #att1, att2, count, AWcount
                    if j < len(sortedComboList[0])-1:
                        f.write(str(sortedComboList[i][j])+'\t')
                    else:
                        f.write(str(sortedComboList[i][j])+'\n')
            f.close()
        else:
            pass


    def save_tracking(self, exploreIter, outFile):
        if cons.doAttributeTracking:
            """ Prints out Attribute Tracking scores to txt file. """
            try:
                f = open(outFile + '_'+ str(exploreIter + 1)+'_AttTrack.txt','w') # Outputs tab delimited text file of rule population and respective rule stats
                print("Writing Attribute Tracking as Data File...")
            except IOError as xxx_todo_changeme4:
                (errno, strerror) = xxx_todo_changeme4.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise

            trackingSums = cons.AT.attAccuracySums
            #-------------------------------------------------------------------
            f.write(str(cons.labelInstanceID) + '\t') #Write InstanceID label
            for att in cons.env.formatData.trainHeaderList:
                f.write(str(att) + '\t')
            f.write(str(cons.labelPhenotype)+ '\n') #Write phenotype label
            #---------------------------------------------------------------
            for i in range(len(trackingSums)):
                trackList = trackingSums[i]
                f.write(str(cons.env.formatData.trainFormatted[i][2])+ '\t') #Write InstanceID
                for att in trackList:
                    f.write(str(att) + '\t')
                f.write(str(cons.env.formatData.trainFormatted[i][1]) +'\n') #Write phenotype

            f.close()


    def writePredictions(self, exploreIter, outFile, predictionList, realList, predictionSets):
        if cons.outputTestPredictions:
            """ Prints out the Testing Predictions to txt file."""
            try:
                f = open(outFile + '_'+ str(exploreIter + 1)+'_Predictions.txt','w') # Outputs tab delimited text file of rule population and respective rule stats
                print("Writing Predictions to File...")
            except IOError as xxx_todo_changeme5:
                (errno, strerror) = xxx_todo_changeme5.args
                print(("I/O error(%s): %s" % (errno, strerror)))
                raise
            f.write(str(cons.labelInstanceID) + '\t'+'Endpoint Predictions'+'\t' + 'True Endpoint')
            if cons.env.formatData.discretePhenotype:
                for eachClass in cons.env.formatData.phenotypeList:
                    f.write('\t'+ str(eachClass))
            f.write('\n')

            for i in range(len(predictionList)):
                f.write(str(cons.env.formatData.testFormatted[i][2])+ '\t') #Write InstanceID
                f.write(str(predictionList[i])+'\t'+str(realList[i]))
                if cons.env.formatData.discretePhenotype:
                    propList = []
                    for eachClass in cons.env.formatData.phenotypeList:
                        propList.append(predictionSets[i][eachClass])
                    for each in propList:
                        f.write('\t'+ str(each))
                f.write('\n')
            f.close()


    def editPopStats(self, testEval):
        """ Takes an existing popStatsFile and edits it to report Testing Accuracy performance, and Testing coverage on a specified testing dataset. """
        try:
            fileObject = open(cons.popRebootPath+"_PopStats.txt", 'rU')  # opens each datafile to read.
        except:
            #break
            print("Data-set Not Found!")
        #Grab the existing file information (only a couple items will change, i.e. testing acccuracy and testing coverage)
        fileStorage = []
        for each in fileObject:
            fileStorage.append(each)
        try:
            fileObject.close()
        except IOError as xxx_todo_changeme6:
            (errno, strerror) = xxx_todo_changeme6.args
            print(("I/O error(%s): %s" % (errno, strerror)))
            raise

        try:
            popStatsOut = open(cons.popRebootPath+'_PopStats_Testing.txt','w') # Outputs Population run stats
            print("Writing Population Statistical Summary File...")
        except IOError as xxx_todo_changeme7:
            (errno, strerror) = xxx_todo_changeme7.args
            print(("I/O error(%s): %s" % (errno, strerror)))
            raise

        for i in range(2):
            popStatsOut.write(fileStorage[i])

        tempList = fileStorage[2].strip().split('\t')
        popStatsOut.write(str(tempList[0])+"\t")
        popStatsOut.write(str(testEval[0])+"\t")
        popStatsOut.write(str(tempList[2]) +"\t")
        popStatsOut.write(str(testEval[1])+"\n\n")

        for i in range(4,36):
            popStatsOut.write(fileStorage[i])

        try:
            popStatsOut.close()
        except IOError as xxx_todo_changeme8:
            (errno, strerror) = xxx_todo_changeme8.args
            print(("I/O error(%s): %s" % (errno, strerror)))
            raise