| | | 1 | | using System.Text; |
| | | 2 | | using System.Text.Json; |
| | | 3 | | using Intentum.Analytics.Models; |
| | | 4 | | using Intentum.Persistence.Repositories; |
| | | 5 | | using Intentum.Runtime.Policy; |
| | | 6 | | |
| | | 7 | | namespace Intentum.Analytics; |
| | | 8 | | |
| | | 9 | | /// <summary> |
| | | 10 | | /// Default implementation of intent analytics using intent history repository. |
| | | 11 | | /// </summary> |
| | | 12 | | public sealed class IntentAnalytics : IIntentAnalytics |
| | | 13 | | { |
| | 1 | 14 | | private static readonly JsonSerializerOptions JsonOptions = new() { WriteIndented = true }; |
| | | 15 | | |
| | | 16 | | private readonly IIntentHistoryRepository _historyRepository; |
| | | 17 | | |
| | 9 | 18 | | public IntentAnalytics(IIntentHistoryRepository historyRepository) |
| | 9 | 19 | | { |
| | 9 | 20 | | _historyRepository = historyRepository ?? throw new ArgumentNullException(nameof(historyRepository)); |
| | 9 | 21 | | } |
| | | 22 | | |
| | | 23 | | /// <inheritdoc /> |
| | | 24 | | public async Task<IReadOnlyList<ConfidenceTrendPoint>> GetConfidenceTrendsAsync( |
| | | 25 | | DateTimeOffset start, |
| | | 26 | | DateTimeOffset end, |
| | | 27 | | TimeSpan bucketSize, |
| | | 28 | | CancellationToken cancellationToken = default) |
| | 2 | 29 | | { |
| | 2 | 30 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 2 | 31 | | var buckets = new Dictionary<(DateTimeOffset Bucket, string Level), (int Count, double Sum)>(); |
| | | 32 | | |
| | 26 | 33 | | foreach (var r in records) |
| | 10 | 34 | | { |
| | 10 | 35 | | var bucketStart = TruncateToBucket(r.RecordedAt, bucketSize); |
| | 10 | 36 | | var key = (bucketStart, r.ConfidenceLevel); |
| | 10 | 37 | | if (!buckets.TryGetValue(key, out var existing)) |
| | 8 | 38 | | existing = (0, 0); |
| | 10 | 39 | | buckets[key] = (existing.Count + 1, existing.Sum + r.ConfidenceScore); |
| | 10 | 40 | | } |
| | | 41 | | |
| | 10 | 42 | | var bucketEnds = buckets.Keys.Select(k => k.Bucket).Distinct().OrderBy(x => x).ToList(); |
| | 10 | 43 | | var levels = buckets.Keys.Select(k => k.Level).Distinct().ToList(); |
| | 2 | 44 | | var result = new List<ConfidenceTrendPoint>(); |
| | | 45 | | |
| | 10 | 46 | | foreach (var bucketStart in bucketEnds) |
| | 2 | 47 | | { |
| | 2 | 48 | | var bucketEnd = bucketStart + bucketSize; |
| | 22 | 49 | | foreach (var level in levels) |
| | 8 | 50 | | { |
| | 8 | 51 | | if (!buckets.TryGetValue((bucketStart, level), out var v)) |
| | 0 | 52 | | continue; |
| | 8 | 53 | | result.Add(new ConfidenceTrendPoint( |
| | 8 | 54 | | bucketStart, |
| | 8 | 55 | | bucketEnd, |
| | 8 | 56 | | level, |
| | 8 | 57 | | v.Count, |
| | 8 | 58 | | v.Count > 0 ? v.Sum / v.Count : 0)); |
| | 8 | 59 | | } |
| | 2 | 60 | | } |
| | | 61 | | |
| | 18 | 62 | | return result.OrderBy(x => x.BucketStart).ThenBy(x => x.ConfidenceLevel).ToList(); |
| | 2 | 63 | | } |
| | | 64 | | |
| | | 65 | | /// <inheritdoc /> |
| | | 66 | | public async Task<DecisionDistributionReport> GetDecisionDistributionAsync( |
| | | 67 | | DateTimeOffset start, |
| | | 68 | | DateTimeOffset end, |
| | | 69 | | CancellationToken cancellationToken = default) |
| | 2 | 70 | | { |
| | 2 | 71 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 2 | 72 | | var countByDecision = records |
| | 10 | 73 | | .GroupBy(r => r.Decision) |
| | 18 | 74 | | .ToDictionary(g => g.Key, g => g.Count()); |
| | | 75 | | |
| | 32 | 76 | | foreach (var decision in Enum.GetValues<PolicyDecision>().Where(d => !countByDecision.ContainsKey(d))) |
| | 6 | 77 | | countByDecision[decision] = 0; |
| | | 78 | | |
| | 2 | 79 | | return new DecisionDistributionReport( |
| | 2 | 80 | | start, |
| | 2 | 81 | | end, |
| | 2 | 82 | | records.Count, |
| | 2 | 83 | | countByDecision); |
| | 2 | 84 | | } |
| | | 85 | | |
| | | 86 | | /// <inheritdoc /> |
| | | 87 | | public async Task<IReadOnlyList<AnomalyReport>> DetectAnomaliesAsync( |
| | | 88 | | DateTimeOffset start, |
| | | 89 | | DateTimeOffset end, |
| | | 90 | | TimeSpan bucketSize, |
| | | 91 | | CancellationToken cancellationToken = default) |
| | 1 | 92 | | { |
| | 1 | 93 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 1 | 94 | | var anomalies = new List<AnomalyReport>(); |
| | | 95 | | |
| | 1 | 96 | | if (!records.Any()) |
| | 0 | 97 | | return anomalies; |
| | | 98 | | |
| | 6 | 99 | | var buckets = records.GroupBy(r => TruncateToBucket(r.RecordedAt, bucketSize)).ToList(); |
| | 2 | 100 | | var bucketCounts = buckets.Select(g => (double)g.Count()).ToList(); |
| | | 101 | | |
| | 1 | 102 | | var (mean, stdDev) = ComputeMeanStdDev(bucketCounts); |
| | 1 | 103 | | var (q1, q3) = ComputeQuartiles(bucketCounts); |
| | 1 | 104 | | var iqr = q3 - q1; |
| | 1 | 105 | | var iqrUpperFence = q3 + 1.5 * iqr; |
| | | 106 | | |
| | 1 | 107 | | var blockRates = buckets.Select(g => |
| | 7 | 108 | | g.Any() ? g.Count(r => r.Decision == PolicyDecision.Block) / (double)g.Count() : 0.0).ToList(); |
| | 1 | 109 | | var (blockMean, blockStdDev) = ComputeMeanStdDev(blockRates); |
| | | 110 | | |
| | 6 | 111 | | foreach (var (g, _) in buckets.Select((g, i) => (g, i))) |
| | 1 | 112 | | { |
| | 1 | 113 | | var bucketStart = g.Key; |
| | 1 | 114 | | var bucketEnd = bucketStart + bucketSize; |
| | 1 | 115 | | var bucketRecords = g.ToList(); |
| | 1 | 116 | | var count = (double)bucketRecords.Count; |
| | | 117 | | |
| | 1 | 118 | | var volumeZScore = stdDev > 0 ? (count - mean) / stdDev : 0; |
| | 1 | 119 | | if (volumeZScore > 2.0 || count > iqrUpperFence) |
| | 0 | 120 | | { |
| | 0 | 121 | | var method = GetVolumeAnomalyMethod(volumeZScore, count, iqrUpperFence); |
| | 0 | 122 | | anomalies.Add(new AnomalyReport( |
| | 0 | 123 | | "VolumeSpike", |
| | 0 | 124 | | $"Volume spike: {bucketRecords.Count} inferences (mean: {mean:F0}, Z: {volumeZScore:F2}, IQR fence: |
| | 0 | 125 | | bucketStart, bucketStart, bucketEnd, |
| | 0 | 126 | | Math.Clamp(volumeZScore / 4, 0, 1), |
| | 0 | 127 | | new Dictionary<string, object> |
| | 0 | 128 | | { |
| | 0 | 129 | | ["Count"] = bucketRecords.Count, ["Mean"] = mean, |
| | 0 | 130 | | ["StdDev"] = stdDev, ["ZScore"] = volumeZScore, |
| | 0 | 131 | | ["IqrUpperFence"] = iqrUpperFence, ["Method"] = method |
| | 0 | 132 | | })); |
| | 0 | 133 | | } |
| | | 134 | | |
| | 1 | 135 | | var blockRate = bucketRecords.Count > 0 |
| | 5 | 136 | | ? bucketRecords.Count(r => r.Decision == PolicyDecision.Block) / (double)bucketRecords.Count |
| | 1 | 137 | | : 0; |
| | 1 | 138 | | var blockZScore = blockStdDev > 0 ? (blockRate - blockMean) / blockStdDev : 0; |
| | 1 | 139 | | if (blockZScore > 2.0) |
| | 0 | 140 | | { |
| | 0 | 141 | | anomalies.Add(new AnomalyReport( |
| | 0 | 142 | | "BlockRateSpike", |
| | 0 | 143 | | $"Block rate spike: {blockRate:P0} (mean: {blockMean:P0}, Z: {blockZScore:F2})", |
| | 0 | 144 | | bucketStart, bucketStart, bucketEnd, |
| | 0 | 145 | | Math.Clamp(blockZScore / 4, 0, 1), |
| | 0 | 146 | | new Dictionary<string, object> |
| | 0 | 147 | | { |
| | 0 | 148 | | ["BlockRate"] = blockRate, ["Mean"] = blockMean, |
| | 0 | 149 | | ["StdDev"] = blockStdDev, ["ZScore"] = blockZScore |
| | 0 | 150 | | })); |
| | 0 | 151 | | } |
| | | 152 | | |
| | 1 | 153 | | var lowConfidenceCount = bucketRecords.Count(r => |
| | 6 | 154 | | string.Equals(r.ConfidenceLevel, "Low", StringComparison.OrdinalIgnoreCase)); |
| | 1 | 155 | | if (bucketRecords.Count >= 2 && lowConfidenceCount / (double)bucketRecords.Count >= 0.5) |
| | 0 | 156 | | { |
| | 0 | 157 | | anomalies.Add(new AnomalyReport( |
| | 0 | 158 | | "LowConfidenceCluster", |
| | 0 | 159 | | $"Low confidence cluster: {lowConfidenceCount}/{bucketRecords.Count} in bucket", |
| | 0 | 160 | | bucketStart, bucketStart, bucketEnd, 0.5, |
| | 0 | 161 | | new Dictionary<string, object> { ["LowCount"] = lowConfidenceCount, ["TotalInBucket"] = bucketRecord |
| | 0 | 162 | | } |
| | 1 | 163 | | } |
| | | 164 | | |
| | 1 | 165 | | return anomalies.OrderByDescending(a => a.Severity).ToList(); |
| | 1 | 166 | | } |
| | | 167 | | |
| | | 168 | | private static (double Mean, double StdDev) ComputeMeanStdDev(List<double> values) |
| | 2 | 169 | | { |
| | 2 | 170 | | if (values.Count == 0) return (0, 0); |
| | 2 | 171 | | var mean = values.Average(); |
| | 4 | 172 | | var variance = values.Sum(v => (v - mean) * (v - mean)) / values.Count; |
| | 2 | 173 | | return (mean, Math.Sqrt(variance)); |
| | 2 | 174 | | } |
| | | 175 | | |
| | | 176 | | private static (double Q1, double Q3) ComputeQuartiles(List<double> values) |
| | 1 | 177 | | { |
| | 2 | 178 | | if (values.Count < 4) return (values.Min(), values.Max()); |
| | 0 | 179 | | var sorted = values.OrderBy(v => v).ToList(); |
| | 0 | 180 | | var n = sorted.Count; |
| | 0 | 181 | | var q1 = sorted[(int)(n * 0.25)]; |
| | 0 | 182 | | var q3 = sorted[(int)(n * 0.75)]; |
| | 0 | 183 | | return (q1, q3); |
| | 1 | 184 | | } |
| | | 185 | | |
| | | 186 | | private static string GetVolumeAnomalyMethod(double volumeZScore, double count, double iqrUpperFence) |
| | 0 | 187 | | { |
| | 0 | 188 | | if (volumeZScore > 2.0 && count > iqrUpperFence) return "Z-score + IQR"; |
| | 0 | 189 | | if (volumeZScore > 2.0) return "Z-score"; |
| | 0 | 190 | | return "IQR"; |
| | 0 | 191 | | } |
| | | 192 | | |
| | | 193 | | /// <inheritdoc /> |
| | | 194 | | public async Task<AnalyticsSummary> GetSummaryAsync( |
| | | 195 | | DateTimeOffset start, |
| | | 196 | | DateTimeOffset end, |
| | | 197 | | TimeSpan bucketSize, |
| | | 198 | | CancellationToken cancellationToken = default) |
| | 1 | 199 | | { |
| | 1 | 200 | | var trends = await GetConfidenceTrendsAsync(start, end, bucketSize, cancellationToken); |
| | 1 | 201 | | var distribution = await GetDecisionDistributionAsync(start, end, cancellationToken); |
| | 1 | 202 | | var anomalies = await DetectAnomaliesAsync(start, end, bucketSize, cancellationToken); |
| | | 203 | | |
| | 1 | 204 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 6 | 205 | | var uniqueSpaces = records.Select(r => r.BehaviorSpaceId).Distinct().Count(); |
| | | 206 | | |
| | 1 | 207 | | return new AnalyticsSummary( |
| | 1 | 208 | | start, |
| | 1 | 209 | | end, |
| | 1 | 210 | | records.Count, |
| | 1 | 211 | | uniqueSpaces, |
| | 1 | 212 | | trends, |
| | 1 | 213 | | distribution, |
| | 1 | 214 | | anomalies); |
| | 1 | 215 | | } |
| | | 216 | | |
| | | 217 | | /// <inheritdoc /> |
| | | 218 | | public async Task<IntentTimeline> GetIntentTimelineAsync( |
| | | 219 | | string entityId, |
| | | 220 | | DateTimeOffset start, |
| | | 221 | | DateTimeOffset end, |
| | | 222 | | CancellationToken cancellationToken = default) |
| | 3 | 223 | | { |
| | 3 | 224 | | var records = await _historyRepository.GetByEntityIdAsync(entityId, start, end, cancellationToken); |
| | 3 | 225 | | var points = records |
| | 4 | 226 | | .Select(r => new IntentTimelinePoint( |
| | 4 | 227 | | r.RecordedAt, |
| | 4 | 228 | | r.IntentName, |
| | 4 | 229 | | r.ConfidenceLevel, |
| | 4 | 230 | | r.ConfidenceScore, |
| | 4 | 231 | | r.Decision)) |
| | 3 | 232 | | .ToList(); |
| | 3 | 233 | | return new IntentTimeline(entityId, start, end, points); |
| | 3 | 234 | | } |
| | | 235 | | |
| | | 236 | | /// <inheritdoc /> |
| | | 237 | | public async Task<IntentGraphSnapshot> GetIntentGraphSnapshotAsync( |
| | | 238 | | string entityId, |
| | | 239 | | DateTimeOffset start, |
| | | 240 | | DateTimeOffset end, |
| | | 241 | | CancellationToken cancellationToken = default) |
| | 1 | 242 | | { |
| | 1 | 243 | | var records = (await _historyRepository.GetByEntityIdAsync(entityId, start, end, cancellationToken)) |
| | 2 | 244 | | .OrderBy(r => r.RecordedAt) |
| | 1 | 245 | | .ToList(); |
| | | 246 | | |
| | 1 | 247 | | var nodes = records |
| | 2 | 248 | | .Select(r => new IntentGraphNode( |
| | 2 | 249 | | r.Id, |
| | 2 | 250 | | r.IntentName, |
| | 2 | 251 | | r.ConfidenceScore, |
| | 2 | 252 | | r.ConfidenceLevel, |
| | 2 | 253 | | r.RecordedAt)) |
| | 1 | 254 | | .ToList(); |
| | | 255 | | |
| | 1 | 256 | | var edges = new List<IntentGraphEdge>(); |
| | 4 | 257 | | for (var i = 0; i < records.Count - 1; i++) |
| | 1 | 258 | | { |
| | 1 | 259 | | var from = records[i]; |
| | 1 | 260 | | var to = records[i + 1]; |
| | 1 | 261 | | edges.Add(new IntentGraphEdge(from.Id, to.Id, to.RecordedAt)); |
| | 1 | 262 | | } |
| | | 263 | | |
| | 1 | 264 | | return new IntentGraphSnapshot( |
| | 1 | 265 | | entityId, |
| | 1 | 266 | | start, |
| | 1 | 267 | | end, |
| | 1 | 268 | | nodes, |
| | 1 | 269 | | edges, |
| | 1 | 270 | | DateTimeOffset.UtcNow); |
| | 1 | 271 | | } |
| | | 272 | | |
| | | 273 | | /// <inheritdoc /> |
| | | 274 | | public async Task<string> ExportToJsonAsync( |
| | | 275 | | DateTimeOffset start, |
| | | 276 | | DateTimeOffset end, |
| | | 277 | | CancellationToken cancellationToken = default) |
| | 1 | 278 | | { |
| | 1 | 279 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 6 | 280 | | var dto = records.Select(r => new |
| | 6 | 281 | | { |
| | 6 | 282 | | r.Id, |
| | 6 | 283 | | r.BehaviorSpaceId, |
| | 6 | 284 | | r.IntentName, |
| | 6 | 285 | | r.ConfidenceLevel, |
| | 6 | 286 | | r.ConfidenceScore, |
| | 6 | 287 | | Decision = r.Decision.ToString(), |
| | 6 | 288 | | r.RecordedAt |
| | 6 | 289 | | }).ToList(); |
| | 1 | 290 | | return JsonSerializer.Serialize(dto, JsonOptions); |
| | 1 | 291 | | } |
| | | 292 | | |
| | | 293 | | /// <inheritdoc /> |
| | | 294 | | public async Task<string> ExportToCsvAsync( |
| | | 295 | | DateTimeOffset start, |
| | | 296 | | DateTimeOffset end, |
| | | 297 | | CancellationToken cancellationToken = default) |
| | 1 | 298 | | { |
| | 1 | 299 | | var records = await _historyRepository.GetByTimeWindowAsync(start, end, cancellationToken); |
| | 1 | 300 | | var sb = new StringBuilder(); |
| | 1 | 301 | | sb.AppendLine("Id,BehaviorSpaceId,IntentName,ConfidenceLevel,ConfidenceScore,Decision,RecordedAt"); |
| | 13 | 302 | | foreach (var r in records) |
| | 5 | 303 | | { |
| | 5 | 304 | | sb.AppendLine($"{EscapeCsv(r.Id)},{EscapeCsv(r.BehaviorSpaceId)},{EscapeCsv(r.IntentName)},{EscapeCsv(r.Conf |
| | 5 | 305 | | } |
| | 1 | 306 | | return sb.ToString(); |
| | 1 | 307 | | } |
| | | 308 | | |
| | | 309 | | private static DateTimeOffset TruncateToBucket(DateTimeOffset value, TimeSpan bucketSize) |
| | 15 | 310 | | { |
| | 15 | 311 | | var ticks = value.UtcTicks - (value.UtcTicks % bucketSize.Ticks); |
| | 15 | 312 | | return new DateTimeOffset(ticks, TimeSpan.Zero); |
| | 15 | 313 | | } |
| | | 314 | | |
| | | 315 | | private static string EscapeCsv(string? value) |
| | 25 | 316 | | { |
| | 25 | 317 | | if (string.IsNullOrEmpty(value)) return ""; |
| | 25 | 318 | | if (value.Contains(',') || value.Contains('"') || value.Contains('\n')) |
| | 0 | 319 | | return "\"" + value.Replace("\"", "\"\"") + "\""; |
| | 25 | 320 | | return value; |
| | 25 | 321 | | } |
| | | 322 | | } |