File size: 62,404 Bytes
824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 f2c26ae 824bf31 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428 429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476 477 478 479 480 481 482 483 484 485 486 487 488 489 490 491 492 493 494 495 496 497 498 499 500 501 502 503 504 505 506 507 508 509 510 511 512 513 514 515 516 517 518 519 520 521 522 523 524 525 526 527 528 529 530 531 532 533 534 535 536 537 538 539 540 541 542 543 544 545 546 547 548 549 550 551 552 553 554 555 556 557 558 559 560 561 562 563 564 565 566 567 568 569 570 571 572 573 574 575 576 577 578 579 580 581 582 583 584 585 586 587 588 589 590 591 592 593 594 595 596 597 598 599 600 601 602 603 604 605 606 607 608 609 610 611 612 613 614 615 616 617 618 619 620 621 622 623 624 625 626 627 628 629 630 631 632 633 634 635 636 637 638 639 640 641 642 643 644 645 646 647 648 649 650 651 652 653 654 655 656 657 658 659 660 661 662 663 664 665 666 667 668 669 670 671 672 673 674 675 676 677 678 679 680 681 682 683 684 685 686 687 688 689 690 691 692 693 694 695 696 697 698 699 700 701 702 703 704 705 706 707 708 709 710 711 712 713 714 715 716 717 718 719 720 721 722 723 724 725 726 727 728 729 730 731 732 733 734 735 736 737 738 739 740 741 742 743 744 745 746 747 748 749 750 751 752 753 754 755 756 757 758 759 760 761 762 763 764 765 766 767 768 769 770 771 772 773 774 775 776 777 778 779 780 781 782 783 784 785 786 787 788 789 790 791 792 793 794 795 796 797 798 799 800 801 802 803 804 805 806 807 808 809 810 811 812 813 814 815 816 817 818 819 820 821 822 823 824 825 826 827 828 829 830 831 832 833 834 835 836 837 838 839 840 841 842 843 844 845 846 847 848 849 850 851 852 853 854 855 856 857 858 859 860 861 862 863 864 865 866 867 868 869 870 871 872 873 874 875 876 877 878 879 880 881 882 883 884 885 886 887 888 889 890 891 892 893 894 895 896 897 898 899 900 901 902 903 904 905 906 907 908 909 910 911 912 913 914 915 916 917 918 919 920 921 922 923 924 925 926 927 928 929 930 931 932 933 934 935 936 937 938 939 940 941 942 943 944 945 946 947 948 949 950 951 952 953 954 955 956 957 958 959 960 961 962 963 964 965 966 967 968 969 970 971 972 973 974 975 976 977 978 979 980 981 982 983 984 985 986 987 988 989 990 991 992 993 994 995 996 997 998 999 1000 1001 1002 1003 1004 1005 1006 1007 1008 1009 1010 1011 1012 1013 1014 1015 1016 1017 1018 1019 1020 1021 1022 1023 1024 1025 1026 1027 1028 1029 1030 1031 1032 1033 1034 1035 1036 1037 1038 1039 1040 1041 1042 1043 1044 1045 1046 1047 1048 1049 1050 1051 1052 1053 1054 1055 1056 1057 1058 1059 1060 1061 1062 1063 1064 1065 1066 1067 1068 1069 1070 1071 1072 1073 1074 1075 1076 1077 1078 1079 1080 1081 1082 1083 1084 1085 1086 1087 1088 1089 1090 1091 1092 1093 1094 1095 1096 1097 1098 1099 1100 1101 1102 1103 1104 1105 1106 1107 1108 1109 1110 1111 1112 1113 1114 1115 1116 1117 1118 1119 1120 1121 1122 1123 1124 1125 1126 1127 1128 1129 1130 1131 1132 1133 1134 1135 1136 1137 1138 1139 1140 1141 1142 1143 1144 1145 1146 1147 1148 1149 1150 1151 1152 1153 1154 1155 1156 1157 1158 1159 1160 1161 1162 1163 1164 1165 1166 1167 1168 1169 1170 1171 1172 1173 1174 1175 1176 1177 1178 1179 1180 1181 1182 1183 1184 1185 1186 1187 1188 1189 1190 1191 1192 1193 1194 1195 1196 1197 1198 1199 1200 1201 1202 1203 1204 1205 1206 1207 1208 1209 1210 1211 1212 1213 1214 1215 1216 1217 1218 1219 1220 1221 1222 1223 1224 1225 1226 1227 1228 1229 1230 1231 1232 1233 1234 1235 1236 1237 1238 1239 1240 1241 1242 1243 1244 1245 1246 1247 1248 1249 1250 1251 1252 1253 1254 1255 1256 1257 1258 1259 1260 1261 1262 1263 1264 1265 1266 1267 1268 1269 1270 1271 1272 1273 1274 1275 1276 1277 1278 1279 1280 1281 1282 1283 1284 1285 1286 1287 1288 1289 1290 1291 1292 1293 1294 1295 1296 1297 1298 1299 1300 1301 1302 1303 1304 1305 1306 1307 1308 1309 1310 1311 1312 1313 1314 1315 1316 1317 1318 1319 1320 1321 1322 1323 1324 1325 1326 1327 1328 1329 1330 1331 1332 1333 1334 1335 1336 1337 1338 1339 1340 1341 1342 1343 1344 1345 1346 1347 1348 1349 1350 1351 1352 1353 1354 1355 1356 1357 1358 1359 1360 1361 1362 1363 1364 1365 1366 1367 1368 1369 1370 1371 1372 1373 1374 1375 1376 1377 1378 1379 1380 1381 1382 1383 1384 1385 1386 1387 1388 1389 1390 1391 1392 1393 1394 1395 1396 1397 1398 1399 1400 1401 1402 1403 1404 1405 1406 |
"""
Module: agents.anita
Codinome: Anita Garibaldi - Roteadora Semântica
Description: Agent specialized in pattern analysis and correlation detection in government data
Author: Anderson H. Silva
Date: 2025-01-24
License: Proprietary - All rights reserved
"""
import asyncio
from datetime import datetime, timedelta
from typing import Any, Dict, List, Optional, Tuple
from dataclasses import dataclass
from collections import defaultdict, Counter
import numpy as np
from pydantic import BaseModel, Field as PydanticField
from src.agents.deodoro import BaseAgent, AgentContext, AgentMessage, AgentResponse
from src.core import get_logger, AgentStatus
from src.core.exceptions import AgentExecutionError, DataAnalysisError
from src.tools.transparency_api import TransparencyAPIClient, TransparencyAPIFilter
from src.ml.spectral_analyzer import SpectralAnalyzer, SpectralFeatures, PeriodicPattern
@dataclass
class PatternResult:
"""Result of pattern analysis."""
pattern_type: str
description: str
significance: float # 0.0 to 1.0
confidence: float # 0.0 to 1.0
insights: List[str]
evidence: Dict[str, Any]
recommendations: List[str]
entities_involved: List[Dict[str, Any]]
trend_direction: Optional[str] = None # "increasing", "decreasing", "stable"
correlation_strength: Optional[float] = None
@dataclass
class CorrelationResult:
"""Result of correlation analysis."""
correlation_type: str
variables: List[str]
correlation_coefficient: float
p_value: Optional[float]
significance_level: str # "high", "medium", "low"
description: str
business_interpretation: str
evidence: Dict[str, Any]
recommendations: List[str]
class AnalysisRequest(BaseModel):
"""Request for pattern and correlation analysis."""
query: str = PydanticField(description="Natural language analysis query")
analysis_types: Optional[List[str]] = PydanticField(default=None, description="Types of analysis to perform")
time_period: Optional[str] = PydanticField(default="12_months", description="Time period for analysis")
organization_codes: Optional[List[str]] = PydanticField(default=None, description="Organizations to analyze")
focus_areas: Optional[List[str]] = PydanticField(default=None, description="Specific areas to focus on")
comparison_mode: bool = PydanticField(default=False, description="Enable comparison between entities")
max_records: int = PydanticField(default=200, description="Maximum records for analysis")
class AnalystAgent(BaseAgent):
"""
Agent specialized in pattern analysis and correlation detection in government data.
Capabilities:
- Spending trend analysis over time
- Organizational spending pattern comparison
- Vendor market behavior analysis
- Seasonal pattern detection
- Contract value distribution analysis
- Cross-organizational correlation analysis
- Performance and efficiency metrics
- Predictive trend modeling
"""
def __init__(
self,
min_correlation_threshold: float = 0.3,
significance_threshold: float = 0.05,
trend_detection_window: int = 6, # months
):
"""
Initialize the Analyst Agent.
Args:
min_correlation_threshold: Minimum correlation coefficient to report
significance_threshold: P-value threshold for statistical significance
trend_detection_window: Number of periods for trend analysis
"""
super().__init__(
name="Anita",
description="Anita Garibaldi - Agent specialized in pattern analysis and correlation detection",
capabilities=[
"spending_trend_analysis",
"organizational_comparison",
"vendor_behavior_analysis",
"seasonal_pattern_detection",
"value_distribution_analysis",
"correlation_analysis",
"efficiency_metrics",
"predictive_modeling"
],
max_retries=3,
timeout=60
)
self.correlation_threshold = min_correlation_threshold
self.significance_threshold = significance_threshold
self.trend_window = trend_detection_window
# Initialize spectral analyzer for frequency-domain analysis
self.spectral_analyzer = SpectralAnalyzer()
# Analysis methods registry
self.analysis_methods = {
"spending_trends": self._analyze_spending_trends,
"organizational_patterns": self._analyze_organizational_patterns,
"vendor_behavior": self._analyze_vendor_behavior,
"seasonal_patterns": self._analyze_seasonal_patterns,
"spectral_patterns": self._analyze_spectral_patterns,
"cross_spectral_analysis": self._perform_cross_spectral_analysis,
"value_distribution": self._analyze_value_distribution,
"correlation_analysis": self._perform_correlation_analysis,
"efficiency_metrics": self._calculate_efficiency_metrics,
}
self.logger.info(
"analyst_agent_initialized",
agent_name=self.name,
correlation_threshold=min_correlation_threshold,
significance_threshold=significance_threshold,
)
async def initialize(self) -> None:
"""Initialize agent resources."""
self.logger.info(f"{self.name} agent initialized")
async def shutdown(self) -> None:
"""Cleanup agent resources."""
self.logger.info(f"{self.name} agent shutting down")
async def process(
self,
message: AgentMessage,
context: AgentContext
) -> AgentResponse:
"""
Process pattern analysis request and return insights.
Args:
message: Analysis request message
context: Agent execution context
Returns:
AgentResponse with patterns and correlations
"""
try:
self.logger.info(
"analysis_started",
investigation_id=context.investigation_id,
agent_name=self.name,
action=message.action,
)
# Parse analysis request
if message.action == "analyze":
request = AnalysisRequest(**message.payload)
else:
raise AgentExecutionError(
f"Unsupported action: {message.action}",
agent_id=self.name
)
# Fetch data for analysis
analysis_data = await self._fetch_analysis_data(request, context)
if not analysis_data:
return AgentResponse(
agent_name=self.name,
status=AgentStatus.COMPLETED,
result={
"status": "no_data",
"message": "No data found for the specified criteria",
"patterns": [],
"correlations": [],
"summary": {"total_records": 0, "patterns_found": 0}
},
metadata={"investigation_id": context.investigation_id}
)
# Perform pattern analysis
patterns = await self._run_pattern_analysis(analysis_data, request, context)
# Perform correlation analysis
correlations = await self._run_correlation_analysis(analysis_data, request, context)
# Generate insights and recommendations
insights = self._generate_insights(patterns, correlations, analysis_data)
# Create result message
result = {
"status": "completed",
"query": request.query,
"patterns": [self._pattern_to_dict(p) for p in patterns],
"correlations": [self._correlation_to_dict(c) for c in correlations],
"insights": insights,
"summary": self._generate_analysis_summary(analysis_data, patterns, correlations),
"metadata": {
"investigation_id": context.investigation_id,
"timestamp": datetime.utcnow().isoformat(),
"agent_name": self.name,
"records_analyzed": len(analysis_data),
"patterns_found": len(patterns),
"correlations_found": len(correlations),
}
}
self.logger.info(
"analysis_completed",
investigation_id=context.investigation_id,
records_analyzed=len(analysis_data),
patterns_found=len(patterns),
correlations_found=len(correlations),
)
return AgentResponse(
agent_name=self.name,
status=AgentStatus.COMPLETED,
result=result,
metadata={"investigation_id": context.investigation_id}
)
except Exception as e:
self.logger.error(
"analysis_failed",
investigation_id=context.investigation_id,
error=str(e),
agent_name=self.name,
)
return AgentResponse(
agent_name=self.name,
status=AgentStatus.ERROR,
error=str(e),
result={
"status": "error",
"error": str(e),
"investigation_id": context.investigation_id,
},
metadata={"investigation_id": context.investigation_id}
)
async def _fetch_analysis_data(
self,
request: AnalysisRequest,
context: AgentContext
) -> List[Dict[str, Any]]:
"""
Fetch comprehensive data for pattern analysis.
Args:
request: Analysis parameters
context: Agent context
Returns:
List of contract records for analysis
"""
all_contracts = []
# Expanded organization codes for broader analysis
org_codes = request.organization_codes or [
"26000", # Ministério da Saúde
"20000", # Presidência da República
"25000", # Ministério da Educação
"36000", # Ministério da Defesa
"44000", # Ministério do Desenvolvimento Social
"30000", # Ministério da Justiça
]
async with TransparencyAPIClient() as client:
for org_code in org_codes:
try:
# Fetch data for multiple months to enable trend analysis
for month in range(1, 13): # Full year
filters = TransparencyAPIFilter(
codigo_orgao=org_code,
ano=2024,
mes=month,
pagina=1,
tamanho_pagina=min(20, request.max_records // (len(org_codes) * 12))
)
response = await client.get_contracts(filters)
# Enrich each contract with metadata
for contract in response.data:
contract["_org_code"] = org_code
contract["_month"] = month
contract["_year"] = 2024
contract["_fetch_timestamp"] = datetime.utcnow().isoformat()
all_contracts.extend(response.data)
# Rate limiting consideration
await asyncio.sleep(0.1)
self.logger.info(
"organization_data_fetched",
org_code=org_code,
total_records=len([c for c in all_contracts if c.get("_org_code") == org_code]),
investigation_id=context.investigation_id,
)
except Exception as e:
self.logger.warning(
"organization_data_fetch_failed",
org_code=org_code,
error=str(e),
investigation_id=context.investigation_id,
)
continue
return all_contracts[:request.max_records]
async def _run_pattern_analysis(
self,
data: List[Dict[str, Any]],
request: AnalysisRequest,
context: AgentContext
) -> List[PatternResult]:
"""
Run pattern analysis algorithms on the data.
Args:
data: Contract records to analyze
request: Analysis parameters
context: Agent context
Returns:
List of detected patterns
"""
all_patterns = []
# Determine which analysis types to run
types_to_run = request.analysis_types or list(self.analysis_methods.keys())
types_to_run = [t for t in types_to_run if t != "correlation_analysis"] # Handle separately
for analysis_type in types_to_run:
if analysis_type in self.analysis_methods:
try:
method = self.analysis_methods[analysis_type]
patterns = await method(data, context)
all_patterns.extend(patterns)
self.logger.info(
"pattern_analysis_completed",
type=analysis_type,
patterns_found=len(patterns),
investigation_id=context.investigation_id,
)
except Exception as e:
self.logger.error(
"pattern_analysis_failed",
type=analysis_type,
error=str(e),
investigation_id=context.investigation_id,
)
# Sort patterns by significance
all_patterns.sort(key=lambda x: x.significance, reverse=True)
return all_patterns
async def _run_correlation_analysis(
self,
data: List[Dict[str, Any]],
request: AnalysisRequest,
context: AgentContext
) -> List[CorrelationResult]:
"""
Run correlation analysis on the data.
Args:
data: Contract records to analyze
request: Analysis parameters
context: Agent context
Returns:
List of detected correlations
"""
correlations = []
if "correlation_analysis" in (request.analysis_types or ["correlation_analysis"]):
try:
correlations = await self._perform_correlation_analysis(data, context)
self.logger.info(
"correlation_analysis_completed",
correlations_found=len(correlations),
investigation_id=context.investigation_id,
)
except Exception as e:
self.logger.error(
"correlation_analysis_failed",
error=str(e),
investigation_id=context.investigation_id,
)
return correlations
async def _analyze_spending_trends(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Analyze spending trends over time."""
patterns = []
# Group spending by month
monthly_spending = defaultdict(float)
monthly_counts = defaultdict(int)
for contract in data:
month = contract.get("_month")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if month and isinstance(valor, (int, float)):
monthly_spending[month] += float(valor)
monthly_counts[month] += 1
if len(monthly_spending) < 3:
return patterns
# Calculate trend
months = sorted(monthly_spending.keys())
values = [monthly_spending[m] for m in months]
# Simple linear regression for trend
x = np.array(range(len(months)))
y = np.array(values)
if len(x) > 1 and np.std(y) > 0:
correlation = np.corrcoef(x, y)[0, 1]
slope = np.polyfit(x, y, 1)[0]
# Determine trend direction and significance
if abs(correlation) > 0.5:
trend_direction = "increasing" if slope > 0 else "decreasing"
significance = abs(correlation)
pattern = PatternResult(
pattern_type="spending_trends",
description=f"Tendência de gastos {trend_direction} detectada",
significance=significance,
confidence=abs(correlation),
insights=[
f"Gastos apresentam tendência {trend_direction} com correlação de {correlation:.2f}",
f"Variação média mensal: R$ {slope:,.2f}",
f"Período analisado: {len(months)} meses",
],
evidence={
"monthly_spending": dict(monthly_spending),
"trend_correlation": correlation,
"monthly_slope": slope,
"total_value": sum(values),
"average_monthly": np.mean(values),
},
recommendations=[
"Investigar fatores que causam a tendência observada",
"Analisar planejamento orçamentário",
"Verificar sazonalidade nos gastos",
"Monitorar sustentabilidade da tendência",
],
entities_involved=[{
"type": "monthly_data",
"months_analyzed": len(months),
"total_contracts": sum(monthly_counts.values()),
}],
trend_direction=trend_direction,
correlation_strength=abs(correlation),
)
patterns.append(pattern)
return patterns
async def _analyze_organizational_patterns(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Analyze spending patterns across organizations."""
patterns = []
# Group by organization
org_stats = defaultdict(lambda: {"total_value": 0, "count": 0, "contracts": []})
for contract in data:
org_code = contract.get("_org_code")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if org_code and isinstance(valor, (int, float)):
org_stats[org_code]["total_value"] += float(valor)
org_stats[org_code]["count"] += 1
org_stats[org_code]["contracts"].append(contract)
if len(org_stats) < 2:
return patterns
# Calculate organization efficiency metrics
org_efficiency = {}
for org_code, stats in org_stats.items():
if stats["count"] > 0:
avg_contract_value = stats["total_value"] / stats["count"]
org_efficiency[org_code] = {
"avg_contract_value": avg_contract_value,
"total_value": stats["total_value"],
"contract_count": stats["count"],
"efficiency_ratio": stats["total_value"] / stats["count"],
}
# Find organizations with unusual patterns
avg_values = [eff["avg_contract_value"] for eff in org_efficiency.values()]
mean_avg = np.mean(avg_values)
std_avg = np.std(avg_values)
for org_code, efficiency in org_efficiency.items():
if std_avg > 0:
z_score = (efficiency["avg_contract_value"] - mean_avg) / std_avg
if abs(z_score) > 1.5: # Significant deviation
pattern_type = "high_value_contracts" if z_score > 0 else "low_value_contracts"
significance = min(abs(z_score) / 3.0, 1.0)
pattern = PatternResult(
pattern_type="organizational_patterns",
description=f"Padrão organizacional atípico: {org_code}",
significance=significance,
confidence=min(abs(z_score) / 2.0, 1.0),
insights=[
f"Organização {org_code} apresenta padrão atípico de contratação",
f"Valor médio por contrato: R$ {efficiency['avg_contract_value']:,.2f}",
f"Desvio da média geral: {z_score:.1f} desvios padrão",
],
evidence={
"organization_code": org_code,
"avg_contract_value": efficiency["avg_contract_value"],
"total_value": efficiency["total_value"],
"contract_count": efficiency["contract_count"],
"z_score": z_score,
"market_average": mean_avg,
},
recommendations=[
"Investigar critérios de contratação da organização",
"Comparar com organizações similares",
"Analisar eficiência dos processos",
"Verificar adequação dos valores contratados",
],
entities_involved=[{
"organization": org_code,
"total_contracts": efficiency["contract_count"],
"total_value": efficiency["total_value"],
}],
)
patterns.append(pattern)
return patterns
async def _analyze_vendor_behavior(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Analyze vendor behavior patterns."""
patterns = []
# Group by vendor
vendor_stats = defaultdict(lambda: {
"contracts": [],
"total_value": 0,
"organizations": set(),
"months": set(),
})
for contract in data:
supplier = contract.get("fornecedor", {})
vendor_name = supplier.get("nome", "Unknown")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
org_code = contract.get("_org_code")
month = contract.get("_month")
if vendor_name != "Unknown" and isinstance(valor, (int, float)):
vendor_stats[vendor_name]["contracts"].append(contract)
vendor_stats[vendor_name]["total_value"] += float(valor)
if org_code:
vendor_stats[vendor_name]["organizations"].add(org_code)
if month:
vendor_stats[vendor_name]["months"].add(month)
# Analyze multi-organization vendors
for vendor_name, stats in vendor_stats.items():
org_count = len(stats["organizations"])
contract_count = len(stats["contracts"])
# Check for vendors working with multiple organizations
if org_count >= 3 and contract_count >= 5:
significance = min(org_count / 6.0, 1.0) # Normalize to max 6 orgs
pattern = PatternResult(
pattern_type="vendor_behavior",
description=f"Fornecedor multi-organizacional: {vendor_name}",
significance=significance,
confidence=min(contract_count / 10.0, 1.0),
insights=[
f"Fornecedor atua em {org_count} organizações diferentes",
f"Total de {contract_count} contratos",
f"Valor total: R$ {stats['total_value']:,.2f}",
f"Presença em {len(stats['months'])} meses diferentes",
],
evidence={
"vendor_name": vendor_name,
"organization_count": org_count,
"contract_count": contract_count,
"total_value": stats["total_value"],
"organizations": list(stats["organizations"]),
"months_active": len(stats["months"]),
},
recommendations=[
"Verificar especialização do fornecedor",
"Analisar competitividade dos processos",
"Investigar relacionamento com múltiplas organizações",
"Revisar histórico de performance",
],
entities_involved=[{
"vendor": vendor_name,
"organizations": list(stats["organizations"]),
"contract_count": contract_count,
}],
)
patterns.append(pattern)
return patterns
async def _analyze_seasonal_patterns(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Analyze seasonal patterns in contracting."""
patterns = []
# Group by month
monthly_activity = defaultdict(lambda: {"count": 0, "value": 0})
for contract in data:
month = contract.get("_month")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if month and isinstance(valor, (int, float)):
monthly_activity[month]["count"] += 1
monthly_activity[month]["value"] += float(valor)
if len(monthly_activity) < 6: # Need at least half year
return patterns
# Calculate monthly averages
months = sorted(monthly_activity.keys())
counts = [monthly_activity[m]["count"] for m in months]
values = [monthly_activity[m]["value"] for m in months]
# Detect end-of-year rush (December spike)
if 12 in monthly_activity and len(months) >= 6:
dec_count = monthly_activity[12]["count"]
avg_count = np.mean([monthly_activity[m]["count"] for m in months if m != 12])
if avg_count > 0:
dec_ratio = dec_count / avg_count
if dec_ratio > 1.5: # 50% above average
significance = min((dec_ratio - 1) / 2, 1.0)
pattern = PatternResult(
pattern_type="seasonal_patterns",
description="Padrão sazonal: concentração em dezembro",
significance=significance,
confidence=min(dec_ratio / 2.0, 1.0),
insights=[
f"Dezembro apresenta {dec_ratio:.1f}x mais contratos que a média",
f"Contratos em dezembro: {dec_count}",
f"Média mensal: {avg_count:.1f}",
"Possível correria de fim de ano orçamentário",
],
evidence={
"december_count": dec_count,
"average_monthly_count": avg_count,
"december_ratio": dec_ratio,
"monthly_distribution": dict(monthly_activity),
},
recommendations=[
"Melhorar planejamento anual de contratações",
"Distribuir contratações ao longo do ano",
"Investigar qualidade dos processos de fim de ano",
"Implementar cronograma de contratações",
],
entities_involved=[{
"pattern": "end_of_year_rush",
"affected_months": [12],
"intensity": dec_ratio,
}],
)
patterns.append(pattern)
return patterns
async def _analyze_value_distribution(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Analyze contract value distribution patterns."""
patterns = []
# Extract contract values
values = []
for contract in data:
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if isinstance(valor, (int, float)) and valor > 0:
values.append(float(valor))
if len(values) < 10:
return patterns
# Calculate distribution statistics
values_array = np.array(values)
# Check for unusual distribution patterns
percentiles = np.percentile(values_array, [25, 50, 75, 90, 95, 99])
# Detect heavy concentration in specific value ranges
value_ranges = {
"micro": (0, 8000), # Dispensas
"small": (8000, 176000), # Convites
"medium": (176000, 1500000), # Tomadas de preço
"large": (1500000, float('inf')) # Concorrências
}
range_counts = {}
range_values = {}
for range_name, (min_val, max_val) in value_ranges.items():
count = sum(1 for v in values if min_val <= v < max_val)
total_val = sum(v for v in values if min_val <= v < max_val)
range_counts[range_name] = count
range_values[range_name] = total_val
total_contracts = len(values)
total_value = sum(values)
# Check for unusual concentrations
for range_name, count in range_counts.items():
percentage = count / total_contracts if total_contracts > 0 else 0
value_percentage = range_values[range_name] / total_value if total_value > 0 else 0
# Detect if one range dominates
if percentage > 0.7: # 70% of contracts in one range
significance = percentage
pattern = PatternResult(
pattern_type="value_distribution",
description=f"Concentração em contratos de valor {range_name}",
significance=significance,
confidence=percentage,
insights=[
f"{percentage:.1%} dos contratos estão na faixa {range_name}",
f"Representam {value_percentage:.1%} do valor total",
f"Total de {count} contratos nesta faixa",
f"Faixa de valores: R$ {value_ranges[range_name][0]:,.2f} - R$ {value_ranges[range_name][1]:,.2f}",
],
evidence={
"range_name": range_name,
"concentration_percentage": percentage * 100,
"value_percentage": value_percentage * 100,
"contract_count": count,
"range_limits": value_ranges[range_name],
"distribution": range_counts,
},
recommendations=[
"Analisar adequação dos valores contratados",
"Verificar se há fracionamento inadequado",
"Revisar modalidades licitatórias utilizadas",
"Comparar com benchmarks do setor",
],
entities_involved=[{
"value_range": range_name,
"contract_count": count,
"percentage": percentage * 100,
}],
)
patterns.append(pattern)
return patterns
async def _perform_correlation_analysis(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[CorrelationResult]:
"""Perform correlation analysis between variables."""
correlations = []
# Prepare data for correlation analysis
# Group by organization and month for time series
org_month_data = defaultdict(lambda: defaultdict(lambda: {"count": 0, "value": 0}))
for contract in data:
org_code = contract.get("_org_code")
month = contract.get("_month")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if org_code and month and isinstance(valor, (int, float)):
org_month_data[org_code][month]["count"] += 1
org_month_data[org_code][month]["value"] += float(valor)
# Analyze correlation between contract count and average value
if len(org_month_data) >= 3:
monthly_counts = []
monthly_avg_values = []
for org_code, month_data in org_month_data.items():
for month, stats in month_data.items():
if stats["count"] > 0:
monthly_counts.append(stats["count"])
monthly_avg_values.append(stats["value"] / stats["count"])
if len(monthly_counts) >= 10 and len(monthly_avg_values) >= 10:
# Calculate correlation between count and average value
correlation_coef = np.corrcoef(monthly_counts, monthly_avg_values)[0, 1]
if abs(correlation_coef) > self.correlation_threshold:
significance_level = "high" if abs(correlation_coef) > 0.7 else "medium"
interpretation = (
"Correlação negativa indica que meses com mais contratos tendem a ter valores médios menores"
if correlation_coef < 0 else
"Correlação positiva indica que meses com mais contratos tendem a ter valores médios maiores"
)
correlation = CorrelationResult(
correlation_type="count_vs_value",
variables=["monthly_contract_count", "monthly_average_value"],
correlation_coefficient=correlation_coef,
p_value=None, # Would need scipy.stats for p-value
significance_level=significance_level,
description=f"Correlação entre quantidade e valor médio de contratos",
business_interpretation=interpretation,
evidence={
"correlation_coefficient": correlation_coef,
"sample_size": len(monthly_counts),
"count_range": [min(monthly_counts), max(monthly_counts)],
"value_range": [min(monthly_avg_values), max(monthly_avg_values)],
},
recommendations=[
"Investigar fatores que influenciam essa correlação",
"Analisar estratégias de contratação",
"Verificar planejamento orçamentário",
"Monitorar tendências futuras",
],
)
correlations.append(correlation)
return correlations
async def _calculate_efficiency_metrics(
self,
data: List[Dict[str, Any]],
context: AgentContext
) -> List[PatternResult]:
"""Calculate efficiency metrics for organizations."""
patterns = []
# Calculate metrics by organization
org_metrics = defaultdict(lambda: {
"total_value": 0,
"contract_count": 0,
"unique_vendors": set(),
"months_active": set(),
})
for contract in data:
org_code = contract.get("_org_code")
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
supplier = contract.get("fornecedor", {}).get("nome")
month = contract.get("_month")
if org_code and isinstance(valor, (int, float)):
org_metrics[org_code]["total_value"] += float(valor)
org_metrics[org_code]["contract_count"] += 1
if supplier:
org_metrics[org_code]["unique_vendors"].add(supplier)
if month:
org_metrics[org_code]["months_active"].add(month)
# Calculate efficiency scores
efficiency_scores = {}
for org_code, metrics in org_metrics.items():
if metrics["contract_count"] > 0:
vendor_diversity = len(metrics["unique_vendors"]) / metrics["contract_count"]
avg_contract_value = metrics["total_value"] / metrics["contract_count"]
activity_consistency = len(metrics["months_active"]) / 12 # Normalize to year
# Composite efficiency score
efficiency_score = (vendor_diversity * 0.4 + activity_consistency * 0.6)
efficiency_scores[org_code] = {
"score": efficiency_score,
"vendor_diversity": vendor_diversity,
"avg_contract_value": avg_contract_value,
"activity_consistency": activity_consistency,
"metrics": metrics,
}
# Find organizations with notably high or low efficiency
if efficiency_scores:
scores = [eff["score"] for eff in efficiency_scores.values()]
mean_score = np.mean(scores)
std_score = np.std(scores)
for org_code, efficiency in efficiency_scores.items():
if std_score > 0:
z_score = (efficiency["score"] - mean_score) / std_score
if abs(z_score) > 1.0: # Significant deviation
performance_type = "high_efficiency" if z_score > 0 else "low_efficiency"
significance = min(abs(z_score) / 2.0, 1.0)
pattern = PatternResult(
pattern_type="efficiency_metrics",
description=f"Performance organizacional {performance_type}: {org_code}",
significance=significance,
confidence=min(abs(z_score) / 1.5, 1.0),
insights=[
f"Score de eficiência: {efficiency['score']:.2f}",
f"Diversidade de fornecedores: {efficiency['vendor_diversity']:.2f}",
f"Consistência de atividade: {efficiency['activity_consistency']:.2f}",
f"Valor médio por contrato: R$ {efficiency['avg_contract_value']:,.2f}",
],
evidence={
"organization": org_code,
"efficiency_score": efficiency["score"],
"vendor_diversity": efficiency["vendor_diversity"],
"activity_consistency": efficiency["activity_consistency"],
"z_score": z_score,
"benchmark_average": mean_score,
},
recommendations=[
"Analisar fatores que contribuem para a performance",
"Compartilhar boas práticas com outras organizações",
"Investigar oportunidades de melhoria" if z_score < 0 else "Manter padrão de excelência",
"Monitorar tendências de performance",
],
entities_involved=[{
"organization": org_code,
"efficiency_score": efficiency["score"],
"performance_type": performance_type,
}],
)
patterns.append(pattern)
return patterns
async def _analyze_spectral_patterns(
self,
data: List[Dict[str, Any]],
request: AnalysisRequest,
context: AgentContext
) -> List[PatternResult]:
"""
Analyze spectral patterns using Fourier transforms.
Args:
data: Contract data for analysis
request: Analysis request parameters
context: Agent context
Returns:
List of spectral pattern results
"""
patterns = []
try:
# Group data by organization for spectral analysis
org_groups = defaultdict(list)
for contract in data:
org_code = contract.get("_org_code", "unknown")
org_groups[org_code].append(contract)
for org_code, org_contracts in org_groups.items():
if len(org_contracts) < 30: # Need sufficient data
continue
# Prepare time series data
time_series_data = self._prepare_time_series_for_org(org_contracts)
if len(time_series_data) < 20:
continue
# Extract spending values and timestamps
spending_data = pd.Series([item['value'] for item in time_series_data])
timestamps = pd.DatetimeIndex([item['date'] for item in time_series_data])
# Perform spectral analysis
spectral_features = self.spectral_analyzer.analyze_time_series(
spending_data, timestamps
)
# Find periodic patterns
periodic_patterns = self.spectral_analyzer.find_periodic_patterns(
spending_data, timestamps, entity_name=f"Org_{org_code}"
)
# Convert to PatternResult objects
for i, period_pattern in enumerate(periodic_patterns[:5]): # Top 5 patterns
if period_pattern.amplitude > 0.1: # Only significant patterns
pattern = PatternResult(
pattern_type="spectral_periodic",
description=f"Padrão periódico detectado: {period_pattern.period_days:.1f} dias",
significance=period_pattern.amplitude,
confidence=period_pattern.confidence,
insights=[
f"Período dominante: {period_pattern.period_days:.1f} dias",
f"Força do padrão: {period_pattern.amplitude:.1%}",
f"Tipo: {period_pattern.pattern_type}",
period_pattern.business_interpretation
],
evidence={
"period_days": period_pattern.period_days,
"frequency_hz": period_pattern.frequency_hz,
"amplitude": period_pattern.amplitude,
"pattern_type": period_pattern.pattern_type,
"confidence": period_pattern.confidence,
"spectral_entropy": spectral_features.spectral_entropy,
"dominant_frequencies": spectral_features.dominant_frequencies,
"seasonal_components": spectral_features.seasonal_components
},
recommendations=[
f"Investigar causa do padrão de {period_pattern.period_days:.1f} dias",
"Verificar se corresponde a processos de negócio conhecidos",
"Analisar se há justificativa administrativa",
"Considerar otimização do cronograma de contratações"
],
entities_involved=[{
"organization_code": org_code,
"contracts_analyzed": len(org_contracts),
"period_days": period_pattern.period_days,
"pattern_strength": period_pattern.amplitude
}],
trend_direction=self._classify_trend_from_spectral(spectral_features),
correlation_strength=period_pattern.amplitude
)
patterns.append(pattern)
# Analyze overall spectral characteristics
if spectral_features.spectral_entropy < 0.3: # Low entropy indicates regular patterns
pattern = PatternResult(
pattern_type="spectral_regularity",
description=f"Padrão de gastos muito regular detectado (entropia: {spectral_features.spectral_entropy:.2f})",
significance=1 - spectral_features.spectral_entropy,
confidence=0.8,
insights=[
f"Entropia espectral baixa: {spectral_features.spectral_entropy:.2f}",
"Gastos seguem padrão muito regular",
"Pode indicar processos automatizados ou planejamento rígido",
f"Anomalia score: {spectral_features.anomaly_score:.2f}"
],
evidence={
"spectral_entropy": spectral_features.spectral_entropy,
"anomaly_score": spectral_features.anomaly_score,
"dominant_frequencies": spectral_features.dominant_frequencies[:5],
"seasonal_components": spectral_features.seasonal_components
},
recommendations=[
"Verificar se a regularidade é justificada",
"Investigar processos de planejamento orçamentário",
"Analisar flexibilidade nos cronogramas",
"Considerar diversificação temporal"
],
entities_involved=[{
"organization_code": org_code,
"spectral_entropy": spectral_features.spectral_entropy,
"regularity_score": 1 - spectral_features.spectral_entropy
}]
)
patterns.append(pattern)
self.logger.info(
"spectral_analysis_completed",
patterns_found=len(patterns),
organizations_analyzed=len(org_groups)
)
except Exception as e:
self.logger.error(f"Error in spectral pattern analysis: {str(e)}")
return patterns
async def _perform_cross_spectral_analysis(
self,
data: List[Dict[str, Any]],
request: AnalysisRequest,
context: AgentContext
) -> List[CorrelationResult]:
"""
Perform cross-spectral analysis between organizations.
Args:
data: Contract data for analysis
request: Analysis request parameters
context: Agent context
Returns:
List of cross-spectral correlation results
"""
correlations = []
try:
# Group data by organization
org_groups = defaultdict(list)
for contract in data:
org_code = contract.get("_org_code", "unknown")
org_groups[org_code].append(contract)
# Get organizations with sufficient data
valid_orgs = {org: contracts for org, contracts in org_groups.items()
if len(contracts) >= 30}
if len(valid_orgs) < 2:
return correlations
org_list = list(valid_orgs.keys())
# Perform pairwise cross-spectral analysis
for i, org1 in enumerate(org_list):
for org2 in org_list[i+1:]:
try:
# Prepare time series for both organizations
ts1 = self._prepare_time_series_for_org(valid_orgs[org1])
ts2 = self._prepare_time_series_for_org(valid_orgs[org2])
if len(ts1) < 20 or len(ts2) < 20:
continue
# Create comparable time series (same date range)
all_dates = sorted(set([item['date'] for item in ts1 + ts2]))
if len(all_dates) < 20:
continue
# Create aligned series
data1 = pd.Series(index=all_dates, dtype=float).fillna(0)
data2 = pd.Series(index=all_dates, dtype=float).fillna(0)
for item in ts1:
data1[item['date']] += item['value']
for item in ts2:
data2[item['date']] += item['value']
timestamps = pd.DatetimeIndex(all_dates)
# Perform cross-spectral analysis
cross_spectral_result = self.spectral_analyzer.cross_spectral_analysis(
data1, data2, f"Org_{org1}", f"Org_{org2}", timestamps
)
if cross_spectral_result and cross_spectral_result.get('max_coherence', 0) > 0.5:
correlation = CorrelationResult(
correlation_type="cross_spectral",
variables=[f"Org_{org1}", f"Org_{org2}"],
correlation_coefficient=cross_spectral_result['correlation_coefficient'],
p_value=None, # Not computed in spectral analysis
significance_level=self._assess_spectral_significance(
cross_spectral_result['max_coherence']
),
description=f"Correlação espectral entre organizações {org1} e {org2}",
business_interpretation=cross_spectral_result['business_interpretation'],
evidence={
"max_coherence": cross_spectral_result['max_coherence'],
"mean_coherence": cross_spectral_result['mean_coherence'],
"correlated_periods_days": cross_spectral_result['correlated_periods_days'],
"synchronization_score": cross_spectral_result['synchronization_score'],
"correlated_frequencies": cross_spectral_result['correlated_frequencies']
},
recommendations=[
"Investigar possível coordenação entre organizações",
"Verificar se há fornecedores em comum",
"Analisar sincronização de processos",
"Revisar independência das contratações"
]
)
correlations.append(correlation)
except Exception as e:
self.logger.warning(f"Cross-spectral analysis failed for {org1}-{org2}: {str(e)}")
continue
self.logger.info(
"cross_spectral_analysis_completed",
correlations_found=len(correlations),
organizations_compared=len(org_list)
)
except Exception as e:
self.logger.error(f"Error in cross-spectral analysis: {str(e)}")
return correlations
def _prepare_time_series_for_org(self, contracts: List[Dict[str, Any]]) -> List[Dict[str, Any]]:
"""Prepare time series data for a specific organization."""
time_series = []
for contract in contracts:
# Extract date
date_str = (
contract.get("dataAssinatura") or
contract.get("dataPublicacao") or
contract.get("dataInicio")
)
if not date_str:
continue
try:
# Parse date (DD/MM/YYYY format)
date_parts = date_str.split("/")
if len(date_parts) == 3:
day, month, year = int(date_parts[0]), int(date_parts[1]), int(date_parts[2])
date_obj = datetime(year, month, day)
# Extract value
valor = contract.get("valorInicial") or contract.get("valorGlobal") or 0
if isinstance(valor, (int, float)) and valor > 0:
time_series.append({
'date': date_obj,
'value': float(valor),
'contract_id': contract.get('id')
})
except (ValueError, IndexError):
continue
# Sort by date and aggregate by date
time_series.sort(key=lambda x: x['date'])
# Aggregate by date
daily_aggregates = defaultdict(float)
for item in time_series:
daily_aggregates[item['date']] += item['value']
return [{'date': date, 'value': value} for date, value in daily_aggregates.items()]
def _classify_trend_from_spectral(self, features: SpectralFeatures) -> Optional[str]:
"""Classify trend direction from spectral features."""
# Analyze trend component
if hasattr(features, 'trend_component') and len(features.trend_component) > 10:
trend_start = np.mean(features.trend_component[:len(features.trend_component)//3])
trend_end = np.mean(features.trend_component[-len(features.trend_component)//3:])
if trend_end > trend_start * 1.1:
return "increasing"
elif trend_end < trend_start * 0.9:
return "decreasing"
else:
return "stable"
return None
def _assess_spectral_significance(self, coherence: float) -> str:
"""Assess significance level of spectral coherence."""
if coherence > 0.8:
return "high"
elif coherence > 0.6:
return "medium"
else:
return "low"
def _generate_insights(
self,
patterns: List[PatternResult],
correlations: List[CorrelationResult],
data: List[Dict[str, Any]]
) -> List[str]:
"""Generate high-level insights from analysis results."""
insights = []
# High-level data insights
total_contracts = len(data)
total_value = sum(
float(c.get("valorInicial") or c.get("valorGlobal") or 0)
for c in data
if isinstance(c.get("valorInicial") or c.get("valorGlobal"), (int, float))
)
insights.append(f"Analisados {total_contracts} contratos totalizando R$ {total_value:,.2f}")
# Pattern insights
if patterns:
high_significance = [p for p in patterns if p.significance > 0.7]
insights.append(f"Identificados {len(patterns)} padrões, sendo {len(high_significance)} de alta significância")
# Most significant pattern
if high_significance:
top_pattern = max(high_significance, key=lambda p: p.significance)
insights.append(f"Padrão mais significativo: {top_pattern.description}")
# Correlation insights
if correlations:
strong_correlations = [c for c in correlations if abs(c.correlation_coefficient) > 0.7]
insights.append(f"Encontradas {len(correlations)} correlações, sendo {len(strong_correlations)} fortes")
# Risk assessment
risk_patterns = [p for p in patterns if p.pattern_type in ["spending_trends", "vendor_behavior"]]
if risk_patterns:
insights.append(f"Identificados {len(risk_patterns)} padrões que requerem atenção especial")
return insights
def _generate_analysis_summary(
self,
data: List[Dict[str, Any]],
patterns: List[PatternResult],
correlations: List[CorrelationResult]
) -> Dict[str, Any]:
"""Generate summary statistics for the analysis."""
# Calculate basic statistics
total_value = sum(
float(c.get("valorInicial") or c.get("valorGlobal") or 0)
for c in data
if isinstance(c.get("valorInicial") or c.get("valorGlobal"), (int, float))
)
organizations = len(set(c.get("_org_code") for c in data if c.get("_org_code")))
months_covered = len(set(c.get("_month") for c in data if c.get("_month")))
# Pattern statistics
pattern_types = Counter(p.pattern_type for p in patterns)
high_significance_patterns = len([p for p in patterns if p.significance > 0.7])
# Calculate overall analysis score
analysis_score = min(
(len(patterns) + len(correlations)) / max(len(data) / 10, 1) * 5,
10
)
return {
"total_records": len(data),
"total_value": total_value,
"organizations_analyzed": organizations,
"months_covered": months_covered,
"patterns_found": len(patterns),
"correlations_found": len(correlations),
"pattern_types": dict(pattern_types),
"high_significance_patterns": high_significance_patterns,
"analysis_score": analysis_score,
"avg_contract_value": total_value / len(data) if data else 0,
}
def _pattern_to_dict(self, pattern: PatternResult) -> Dict[str, Any]:
"""Convert PatternResult to dictionary for serialization."""
return {
"type": pattern.pattern_type,
"description": pattern.description,
"significance": pattern.significance,
"confidence": pattern.confidence,
"insights": pattern.insights,
"evidence": pattern.evidence,
"recommendations": pattern.recommendations,
"entities_involved": pattern.entities_involved,
"trend_direction": pattern.trend_direction,
"correlation_strength": pattern.correlation_strength,
}
def _correlation_to_dict(self, correlation: CorrelationResult) -> Dict[str, Any]:
"""Convert CorrelationResult to dictionary for serialization."""
return {
"type": correlation.correlation_type,
"variables": correlation.variables,
"correlation_coefficient": correlation.correlation_coefficient,
"p_value": correlation.p_value,
"significance_level": correlation.significance_level,
"description": correlation.description,
"business_interpretation": correlation.business_interpretation,
"evidence": correlation.evidence,
"recommendations": correlation.recommendations,
} |