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