[518] | 1 | package de.ugoe.cs.quest.coverage;
|
---|
| 2 |
|
---|
| 3 | import java.security.InvalidParameterException;
|
---|
| 4 | import java.util.Collection;
|
---|
| 5 | import java.util.LinkedHashSet;
|
---|
| 6 | import java.util.List;
|
---|
| 7 | import java.util.Map;
|
---|
| 8 |
|
---|
| 9 | import de.ugoe.cs.quest.eventcore.Event;
|
---|
| 10 | import de.ugoe.cs.quest.usageprofiles.IStochasticProcess;
|
---|
| 11 |
|
---|
| 12 | /**
|
---|
| 13 | * <p>
|
---|
| 14 | * This class calculates various types of sequence coverage in relation to a
|
---|
| 15 | * collection of observed sequences.
|
---|
| 16 | * </p>
|
---|
| 17 | *
|
---|
| 18 | * @author Steffen Herbold
|
---|
| 19 | * @version 1.0
|
---|
| 20 | */
|
---|
| 21 | public class CoverageCalculatorObserved {
|
---|
| 22 |
|
---|
| 23 | /**
|
---|
| 24 | * <p>
|
---|
| 25 | * Sequences for which the coverage is calculated.
|
---|
| 26 | * </p>
|
---|
| 27 | */
|
---|
[547] | 28 | private final Collection<List<Event>> sequences;
|
---|
[518] | 29 |
|
---|
| 30 | /**
|
---|
| 31 | * <p>
|
---|
| 32 | * Observed sequences that are baseline for the coverage calculation.
|
---|
| 33 | * </p>
|
---|
| 34 | */
|
---|
[547] | 35 | private final Collection<List<Event>> observedSequences;
|
---|
[518] | 36 |
|
---|
| 37 | /**
|
---|
| 38 | * <p>
|
---|
| 39 | * Length of the subsequences in relation to which the covarage is
|
---|
| 40 | * calculated.
|
---|
| 41 | * </p>
|
---|
| 42 | */
|
---|
| 43 | private final int length;
|
---|
| 44 |
|
---|
| 45 | /**
|
---|
| 46 | * <p>
|
---|
| 47 | * All subsequences of {@link #length} of {@link #sequences}.
|
---|
| 48 | * </p>
|
---|
| 49 | */
|
---|
[547] | 50 | private Collection<List<Event>> subSeqsGenerated = null;
|
---|
[518] | 51 |
|
---|
| 52 | /**
|
---|
| 53 | * <p>
|
---|
| 54 | * All subsequences of {@link #length} of {@link #observedSequences}.
|
---|
| 55 | * </p>
|
---|
| 56 | */
|
---|
[547] | 57 | private Collection<List<Event>> subSeqsObserved = null;
|
---|
[518] | 58 |
|
---|
| 59 | /**
|
---|
| 60 | * <p>
|
---|
| 61 | * Constructor. Creates a new CoverageCalculatorObserved for given
|
---|
| 62 | * collections of observed sequences and generated sequences.
|
---|
| 63 | * </p>
|
---|
| 64 | *
|
---|
| 65 | * @param observedSequences
|
---|
| 66 | * observed sequences in relation to which the coverage is
|
---|
| 67 | * calculated; must not be null
|
---|
| 68 | * @param sequences
|
---|
| 69 | * sequences for which the coverage is calculated; must not be
|
---|
| 70 | * null
|
---|
| 71 | * @param length
|
---|
| 72 | * length of the subsequences for which the coverage is analyzed;
|
---|
| 73 | * must be >0
|
---|
| 74 | * @throws InvalidParameterException
|
---|
| 75 | * thrown if observedSequences or sequences is null or length
|
---|
| 76 | * less than or equal to 0
|
---|
| 77 | */
|
---|
| 78 | public CoverageCalculatorObserved(
|
---|
[547] | 79 | Collection<List<Event>> observedSequences,
|
---|
| 80 | Collection<List<Event>> sequences, int length) {
|
---|
[518] | 81 | if (observedSequences == null) {
|
---|
| 82 | throw new InvalidParameterException(
|
---|
| 83 | "observed sequences must not be null");
|
---|
| 84 | }
|
---|
| 85 | if (sequences == null) {
|
---|
| 86 | throw new InvalidParameterException("sequences must not be null");
|
---|
| 87 | }
|
---|
| 88 | if (length <= 0) {
|
---|
| 89 | throw new InvalidParameterException(
|
---|
| 90 | "length must be >0; actual value: " + length);
|
---|
| 91 | }
|
---|
| 92 | this.observedSequences = observedSequences;
|
---|
| 93 | this.sequences = sequences;
|
---|
| 94 | this.length = length;
|
---|
| 95 | }
|
---|
| 96 |
|
---|
| 97 | /**
|
---|
| 98 | * <p>
|
---|
| 99 | * Calculates the percentage of subsequences of length k that occur, with
|
---|
| 100 | * reference to those that were observed.
|
---|
| 101 | * </p>
|
---|
| 102 | *
|
---|
| 103 | * @return coverage percentage
|
---|
| 104 | */
|
---|
| 105 | public double getCoverageObserved() {
|
---|
| 106 | createSubSeqs();
|
---|
[547] | 107 | Collection<List<Event>> subSeqsObservedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 108 | subSeqsObserved);
|
---|
| 109 | subSeqsObservedCopy.retainAll(subSeqsGenerated);
|
---|
| 110 | return ((double) subSeqsObservedCopy.size()) / subSeqsObserved.size();
|
---|
| 111 | }
|
---|
| 112 |
|
---|
| 113 | /**
|
---|
| 114 | * <p>
|
---|
| 115 | * Calculates the weight of subsequences of length k that occur, with
|
---|
| 116 | * reference to those that were observed.
|
---|
| 117 | * </p>
|
---|
| 118 | *
|
---|
| 119 | * @param process
|
---|
| 120 | * stochastic process in reference to which the weight is
|
---|
| 121 | * calculated
|
---|
| 122 | * @return coverage percentage
|
---|
| 123 | */
|
---|
| 124 |
|
---|
| 125 | public double getCoverageObservedWeigth(IStochasticProcess process) {
|
---|
| 126 | createSubSeqs();
|
---|
[547] | 127 | Map<List<Event>, Double> weightMap = SequenceTools
|
---|
[518] | 128 | .generateWeights(process, subSeqsObserved);
|
---|
| 129 |
|
---|
[547] | 130 | Collection<List<Event>> subSeqsObservedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 131 | subSeqsObserved);
|
---|
| 132 | subSeqsObservedCopy.retainAll(subSeqsGenerated);
|
---|
| 133 | double weight = 0.0d;
|
---|
[547] | 134 | for (List<Event> subSeq : subSeqsObservedCopy) {
|
---|
[518] | 135 | weight += weightMap.get(subSeq);
|
---|
| 136 | }
|
---|
| 137 | return weight;
|
---|
| 138 | }
|
---|
| 139 |
|
---|
| 140 | /**
|
---|
| 141 | * <p>
|
---|
| 142 | * Calculates the percentage of generated subsequences of length k that
|
---|
| 143 | * occur and have not been observed, with reference to all generated
|
---|
| 144 | * subsequences.
|
---|
| 145 | * </p>
|
---|
| 146 | *
|
---|
| 147 | * @return coverage percentage
|
---|
| 148 | */
|
---|
| 149 | public double getNewPercentage() {
|
---|
| 150 | createSubSeqs();
|
---|
[547] | 151 | Collection<List<Event>> subSeqsGeneratedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 152 | subSeqsGenerated);
|
---|
| 153 | subSeqsGeneratedCopy.removeAll(subSeqsObserved);
|
---|
| 154 | return ((double) subSeqsGeneratedCopy.size()) / subSeqsGenerated.size();
|
---|
| 155 | }
|
---|
| 156 |
|
---|
| 157 | /**
|
---|
| 158 | * <p>
|
---|
| 159 | * Calculates the percentage of generated subsequences of length k that
|
---|
| 160 | * occur and have not been observed, with references to all possible new
|
---|
| 161 | * subsequences.
|
---|
| 162 | * </p>
|
---|
| 163 | *
|
---|
| 164 | * @param process
|
---|
| 165 | * stochastic process which is used to determine which
|
---|
| 166 | * subsequences are possible
|
---|
| 167 | * @return coverage percentage
|
---|
| 168 | * @throws InvalidParameterException
|
---|
| 169 | * thrown if process is null
|
---|
| 170 | */
|
---|
| 171 | public double getCoveragePossibleNew(IStochasticProcess process) {
|
---|
| 172 | if (process == null) {
|
---|
| 173 | throw new InvalidParameterException("process must not be null");
|
---|
| 174 | }
|
---|
| 175 | createSubSeqs();
|
---|
[547] | 176 | Collection<List<Event>> subSeqsGeneratedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 177 | subSeqsGenerated);
|
---|
[547] | 178 | Collection<List<Event>> subSeqsPossible = process
|
---|
[518] | 179 | .generateSequences(length);
|
---|
| 180 | subSeqsGeneratedCopy.removeAll(subSeqsObserved);
|
---|
| 181 | subSeqsPossible.removeAll(subSeqsObserved);
|
---|
| 182 | int possibleSize = subSeqsPossible.size();
|
---|
| 183 | subSeqsPossible.retainAll(subSeqsGeneratedCopy);
|
---|
| 184 | return ((double) subSeqsPossible.size()) / possibleSize;
|
---|
| 185 | }
|
---|
| 186 |
|
---|
| 187 | /**
|
---|
| 188 | * <p>
|
---|
| 189 | * Calculates the weight of generated subsequences of length k that occur
|
---|
| 190 | * and have not been observed, with references to all possible new
|
---|
| 191 | * subsequences.
|
---|
| 192 | * </p>
|
---|
| 193 | *
|
---|
| 194 | * @param process
|
---|
| 195 | * stochastic process which is used to determine the weights and
|
---|
| 196 | * which subsequences are possible
|
---|
| 197 | * @return coverage percentage
|
---|
| 198 | * @throws InvalidParameterException
|
---|
| 199 | * thrown if process is null
|
---|
| 200 | */
|
---|
| 201 | public double getCoveragePossibleNewWeight(IStochasticProcess process) {
|
---|
| 202 | if (process == null) {
|
---|
| 203 | throw new InvalidParameterException("process must not be null");
|
---|
| 204 | }
|
---|
| 205 | createSubSeqs();
|
---|
[547] | 206 | Collection<List<Event>> subSeqsGeneratedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 207 | subSeqsGenerated);
|
---|
[547] | 208 | Collection<List<Event>> subSeqsPossible = process
|
---|
[518] | 209 | .generateSequences(length);
|
---|
| 210 | subSeqsGeneratedCopy.removeAll(subSeqsObserved);
|
---|
| 211 | subSeqsPossible.removeAll(subSeqsObserved);
|
---|
[547] | 212 | Map<List<Event>, Double> weightMap = SequenceTools
|
---|
[518] | 213 | .generateWeights(process, subSeqsPossible);
|
---|
| 214 | double weight = 0.0d;
|
---|
[547] | 215 | for (List<Event> subSeq : subSeqsGeneratedCopy) {
|
---|
[518] | 216 | Double currentWeight = weightMap.get(subSeq);
|
---|
| 217 | if( currentWeight!=null ) {
|
---|
| 218 | weight += currentWeight;
|
---|
| 219 | }
|
---|
| 220 | }
|
---|
| 221 | return weight;
|
---|
| 222 | }
|
---|
| 223 |
|
---|
| 224 | /**
|
---|
| 225 | * <p>
|
---|
| 226 | * Returns the number of covered subsequences of length k.
|
---|
| 227 | * </p>
|
---|
| 228 | *
|
---|
| 229 | * @return number of covered subsequences
|
---|
| 230 | */
|
---|
| 231 | public int getNumObserved() {
|
---|
| 232 | createSubSeqs();
|
---|
| 233 | return subSeqsObserved.size();
|
---|
| 234 | }
|
---|
| 235 |
|
---|
| 236 | /**
|
---|
| 237 | * <p>
|
---|
| 238 | * Returns the number of covered subsequences of length k.
|
---|
| 239 | * </p>
|
---|
| 240 | *
|
---|
| 241 | * @return number of covered subsequences
|
---|
| 242 | */
|
---|
| 243 | public int getNumCovered() {
|
---|
| 244 | createSubSeqs();
|
---|
| 245 | return subSeqsGenerated.size();
|
---|
| 246 | }
|
---|
| 247 |
|
---|
| 248 | public int getNumNew() {
|
---|
| 249 | createSubSeqs();
|
---|
[547] | 250 | Collection<List<Event>> subSeqsGeneratedCopy = new LinkedHashSet<List<Event>>(
|
---|
[518] | 251 | subSeqsGenerated);
|
---|
| 252 | subSeqsGeneratedCopy.removeAll(subSeqsObserved);
|
---|
| 253 | return subSeqsGeneratedCopy.size();
|
---|
| 254 | }
|
---|
| 255 |
|
---|
| 256 | /**
|
---|
| 257 | * <p>
|
---|
| 258 | * Helper function that calcuates the subsequences of length k that have
|
---|
| 259 | * been observed and generated.
|
---|
| 260 | * </p>
|
---|
| 261 | */
|
---|
| 262 | private void createSubSeqs() {
|
---|
| 263 | if (subSeqsObserved == null) {
|
---|
| 264 | subSeqsObserved = SequenceTools.containedSubSequences(
|
---|
| 265 | observedSequences, length);
|
---|
| 266 | }
|
---|
| 267 | if (subSeqsGenerated == null) {
|
---|
| 268 | subSeqsGenerated = SequenceTools.containedSubSequences(sequences,
|
---|
| 269 | length);
|
---|
| 270 | }
|
---|
| 271 | }
|
---|
| 272 | }
|
---|