db-changelog-1.51.xml

<?xml version="1.0" encoding="UTF-8"?>
<databaseChangeLog
xmlns="http://www.liquibase.org/xml/ns/dbchangelog"
xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"
xsi:schemaLocation="http://www.liquibase.org/xml/ns/dbchangelog
                      http://www.liquibase.org/xml/ns/dbchangelog/dbchangelog-3.6.xsd">

<!-- 
  Historical Election Cycle Trend Views - Database Changelog v1.51 (REVISED)
  Author: Intelligence Operative & Political Analyst
  Date: 2026-01-13 (Revised: 2026-01-13)
  GitHub Issue: #8205 - Historical Election Cycle Trend Views: Swedish Parliament Context
  Related PR: #8204 - Framework-validation sample data
  
  REVISION NOTES:
  This is a comprehensive revision that properly integrates ALL advanced views
  for true META/META-level analysis across all 6 analytical frameworks.
  
  Key Improvements:
  1. Temporal Analysis: Added view_decision_temporal_trends, view_committee_productivity
  2. Comparative Analysis: Added view_party_performance_metrics, view_committee_productivity_matrix
  3. Pattern Recognition: Added view_riksdagen_voting_anomaly_detection, view_politician_risk_summary
  4. Predictive Intelligence: Added view_ministry_risk_evolution, view_party_effectiveness_trends
  5. Network Analysis: Added view_riksdagen_politician_influence_metrics
  6. Decision Intelligence: Added view_decision_temporal_trends, view_ministry_decision_impact
  
  Purpose:
  Creates 6 comprehensive META/META-level database views for Swedish Parliament election 
  cycle analysis, covering ALL 6 analytical frameworks with Swedish parliamentary context.
  
  Swedish Parliamentary Context:
  - Elections every 4 years (since 1994) on second Sunday in September
  - Parliamentary periods start second Tuesday in September
  - Autumn semester: mid-September to ~January 25
  - Spring semester: ~January 26 to next September
  - Final spring semester before election is high-significance (campaign impact, policy shifts)
  
  Key Design Principles:
  1. META/META Level: Only aggregate existing advanced views, never query base tables
  2. Swedish Context: 4-year cycles, autumn/spring semesters, pre-election markers
  3. Framework Coverage: All 6 analytical frameworks explicitly addressed
  4. Explicit Mapping: Every column mapped to source view with SQL comments
  5. Temporal Dimensions: election_cycle_id, year, semester, pre-election significance
  
  Views Created:
  1. view_election_cycle_temporal_trends (Temporal Analysis framework - 35 supporting views)
  2. view_election_cycle_comparative_analysis (Comparative Analysis - 26 supporting views)
  3. view_election_cycle_anomaly_pattern (Pattern Recognition - 23 supporting views)
  4. view_election_cycle_predictive_intelligence (Predictive Intelligence - 14 supporting views)
  5. view_election_cycle_network_analysis (Network Analysis - 11 supporting views)
  6. view_election_cycle_decision_intelligence (Decision Intelligence - 5 supporting views)
  
  Intelligence Applications:
  - Election-aware political scorecards
  - Coalition stability analysis across election cycles
  - Pre-election behavioral pattern detection
  - Cross-cycle comparative analysis
  - Predictive election forecasting
  - Legislative effectiveness by cycle phase
-->

<changeSet id="1.51-intro" author="intelligence-operative">
    <comment>
        Database Changelog v1.51 - Historical Election Cycle Trend Views (REVISED)
        
        Creates 6 META/META-level views aggregating existing advanced analytics
        with Swedish parliamentary election cycle temporal dimensions.
        
        REVISION: Now properly integrates ALL relevant advanced views for each framework
        to provide comprehensive, multi-source intelligence aggregation.
        
        Enables institutional-grade, election-season-aware intelligence across
        all 6 analytical frameworks (Temporal, Comparative, Pattern Recognition,
        Predictive, Network, Decision Intelligence).
    </comment>
    
    <sql>
        SELECT 
            'v1.51-revised' AS version,
            'Election cycle trend views - COMPREHENSIVE META/META level with all advanced views' AS description,
            'Covers 6 frameworks: Temporal, Comparative, Pattern, Predictive, Network, Decision' AS scope,
            CURRENT_TIMESTAMP AS applied_at;
    </sql>
</changeSet>


<!-- ========================================================================= -->
<!-- 1. TEMPORAL ANALYSIS (ENHANCED): view_election_cycle_temporal_trends    -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-temporal-001" failOnError="true">
    <comment>
        REVISED: Enhanced with view_decision_temporal_trends and view_committee_productivity
        
        Framework: Temporal Analysis (35 supporting views, 20+ risk rules)
        
        Source Views (COMPREHENSIVE META level):
        - view_politician_behavioral_trends: Behavioral metrics, moving averages
        - view_riksdagen_vote_data_ballot_politician_summary_monthly: Voting patterns
        - view_decision_temporal_trends: Decision outcomes, approval rates (NEW)
        - view_committee_productivity: Committee activity metrics (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_temporal_trends">
        <![CDATA[
WITH election_cycle_calendar AS (
    SELECT 
        year_series AS calendar_year,
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4))::INTEGER AS cycle_start_year,
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4)::INTEGER AS cycle_end_year
    FROM generate_series(1994, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
),
election_cycle_periods AS (
    SELECT 
        calendar_year,
        cycle_start_year,
        cycle_end_year,
        (cycle_start_year || '-' || cycle_end_year) AS election_cycle_id,
        (calendar_year - cycle_start_year + 1) AS cycle_year,
        MAKE_DATE(calendar_year, 9, 1) AS autumn_start,
        MAKE_DATE(calendar_year + 1, 1, 25) AS autumn_end,
        MAKE_DATE(calendar_year, 1, 26) AS spring_start,
        MAKE_DATE(calendar_year, 8, 31) AS spring_end,
        CASE WHEN (calendar_year - cycle_start_year + 1) = 4 THEN TRUE ELSE FALSE END AS is_election_year
    FROM election_cycle_calendar
),
temporal_data_autumn AS (
    SELECT 
        ecp.election_cycle_id,
        ecp.cycle_year,
        ecp.calendar_year,
        'autumn' AS semester,
        FALSE AS is_pre_election_semester,
        EXTRACT(YEAR FROM AGE(MAKE_DATE(ecp.cycle_end_year, 9, 15), ecp.autumn_start))::INTEGER * 12 + 
        EXTRACT(MONTH FROM AGE(MAKE_DATE(ecp.cycle_end_year, 9, 15), ecp.autumn_start))::INTEGER AS months_until_election,
        
        -- From view_politician_behavioral_trends
        COUNT(DISTINCT pbt.person_id) AS active_politicians,
        ROUND(AVG(100 - pbt.avg_absence_rate), 2) AS avg_attendance_rate,
        SUM(pbt.total_ballots) AS total_ballots,
        SUM(pbt.total_votes) AS total_votes,
        ROUND(AVG(pbt.avg_win_rate), 2) AS avg_win_rate,
        ROUND(AVG(pbt.avg_rebel_rate), 2) AS avg_rebel_rate,
        SUM(pbt.violation_count) AS violation_count,
        ROUND(AVG(pbt.ma_3month_absence), 2) AS avg_ma_absence,
        
        -- From view_decision_temporal_trends (NEW)
        ROUND(AVG(dtt.daily_approval_rate), 2) AS avg_approval_rate,
        SUM(dtt.daily_decisions) AS total_decisions,
        
        -- From view_committee_productivity (NEW)
        ROUND(AVG(cp.productivity_score), 2) AS avg_committee_productivity
        
    FROM election_cycle_periods ecp
    LEFT JOIN view_politician_behavioral_trends pbt
        ON pbt.period_start >= ecp.autumn_start AND pbt.period_start <= ecp.autumn_end
    LEFT JOIN view_decision_temporal_trends dtt
        ON dtt.decision_day >= ecp.autumn_start AND dtt.decision_day <= ecp.autumn_end
    LEFT JOIN view_committee_productivity cp
        ON 1=1  -- Current snapshot, no temporal filtering available
    GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, ecp.autumn_start, ecp.cycle_end_year
),
temporal_data_spring AS (
    SELECT 
        ecp.election_cycle_id,
        ecp.cycle_year,
        ecp.calendar_year,
        'spring' AS semester,
        CASE WHEN ecp.is_election_year THEN TRUE ELSE FALSE END AS is_pre_election_semester,
        EXTRACT(YEAR FROM AGE(MAKE_DATE(ecp.cycle_end_year, 9, 15), ecp.spring_start))::INTEGER * 12 + 
        EXTRACT(MONTH FROM AGE(MAKE_DATE(ecp.cycle_end_year, 9, 15), ecp.spring_start))::INTEGER AS months_until_election,
        
        COUNT(DISTINCT pbt.person_id) AS active_politicians,
        ROUND(AVG(100 - pbt.avg_absence_rate), 2) AS avg_attendance_rate,
        SUM(pbt.total_ballots) AS total_ballots,
        SUM(pbt.total_votes) AS total_votes,
        ROUND(AVG(pbt.avg_win_rate), 2) AS avg_win_rate,
        ROUND(AVG(pbt.avg_rebel_rate), 2) AS avg_rebel_rate,
        SUM(pbt.violation_count) AS violation_count,
        ROUND(AVG(pbt.ma_3month_absence), 2) AS avg_ma_absence,
        ROUND(AVG(dtt.daily_approval_rate), 2) AS avg_approval_rate,
        SUM(dtt.daily_decisions) AS total_decisions,
        ROUND(AVG(cp.productivity_score), 2) AS avg_committee_productivity
        
    FROM election_cycle_periods ecp
    LEFT JOIN view_politician_behavioral_trends pbt
        ON pbt.period_start >= ecp.spring_start AND pbt.period_start <= ecp.spring_end
    LEFT JOIN view_decision_temporal_trends dtt
        ON dtt.decision_day >= ecp.spring_start AND dtt.decision_day <= ecp.spring_end
    LEFT JOIN view_committee_productivity cp
        ON 1=1  -- Current snapshot, no temporal filtering available
    GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, ecp.spring_start, ecp.cycle_end_year, ecp.is_election_year
)
SELECT * FROM temporal_data_autumn
UNION ALL
SELECT * FROM temporal_data_spring
ORDER BY election_cycle_id, cycle_year, semester;
        ]]>
    </createView>
    
    <rollback>
        <dropView viewName="view_election_cycle_temporal_trends"/>
    </rollback>
</changeSet>


<!-- ========================================================================= -->
<!-- 2. COMPARATIVE ANALYSIS (ENHANCED): view_election_cycle_comparative_analysis -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-comparative-002" failOnError="true">
    <comment>
        REVISED: Enhanced with view_party_performance_metrics and view_committee_productivity_matrix
        
        Framework: Comparative Analysis (26 supporting views, 15+ risk rules)
        
        Source Views (COMPREHENSIVE):
        - view_riksdagen_vote_data_ballot_party_summary_monthly: Party voting patterns
        - view_party_performance_metrics: Performance scores, discipline, effectiveness (NEW)
        - view_committee_productivity_matrix: Multi-committee comparisons (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_comparative_analysis">
        <![CDATA[
WITH election_cycle_periods AS (
    SELECT 
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) || '-' || (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4) AS election_cycle_id,
        (year_series - (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) + 1) AS cycle_year,
        year_series AS calendar_year
    FROM generate_series(1994, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
)
SELECT 
    ecp.election_cycle_id,
    ecp.cycle_year,
    ecp.calendar_year,
    'annual' AS semester,  -- Simplified: aggregate by year since source views are current snapshots
    ppm.party,
    
    -- From view_party_performance_metrics (NEW - provides comprehensive party analytics)
    MAX(ppm.performance_score) AS performance_score,
    MAX(ppm.active_members) AS active_members,
    MAX(ppm.total_violations) AS party_violations,
    ROUND(AVG(ppm.avg_win_rate), 2) AS party_win_rate,
    ROUND(AVG(ppm.avg_participation_rate), 2) AS party_participation_rate,
    MAX(ppm.documents_last_year) AS documents_last_year,
    ROUND(AVG(ppm.avg_rebel_rate), 2) AS party_avg_rebel_rate,
    
    -- From view_committee_productivity_matrix (NEW - multi-committee dimension)
    COUNT(DISTINCT cpm.committee_code) AS committees_active
    
FROM election_cycle_periods ecp
CROSS JOIN view_party_performance_metrics ppm
LEFT JOIN view_committee_productivity_matrix cpm
    ON EXTRACT(YEAR FROM cpm.period_start) = ecp.calendar_year
WHERE ppm.party IS NOT NULL
GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, ppm.party
ORDER BY election_cycle_id, cycle_year, semester, party;
        ]]>
    </createView>
    <rollback><dropView viewName="view_election_cycle_comparative_analysis"/></rollback>
</changeSet>


<!-- ========================================================================= -->
<!-- 3. PATTERN RECOGNITION (ENHANCED): view_election_cycle_anomaly_pattern  -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-pattern-003" failOnError="true">
    <comment>
        REVISED: Enhanced with view_riksdagen_voting_anomaly_detection and view_politician_risk_summary
        
        Framework: Pattern Recognition (23 supporting views, 12/13 risk rules)
        
        Source Views (COMPREHENSIVE):
        - view_risk_score_evolution: Risk trajectories and escalations
        - view_riksdagen_voting_anomaly_detection: Defection risk, discipline issues (NEW)
        - view_politician_risk_summary: Aggregated risk profiles (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_anomaly_pattern">
        <![CDATA[
WITH election_cycle_periods AS (
    SELECT 
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) || '-' || (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4) AS election_cycle_id,
        (year_series - (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) + 1) AS cycle_year,
        year_series AS calendar_year
    FROM generate_series(1994, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
)
SELECT 
    ecp.election_cycle_id,
    ecp.cycle_year,
    ecp.calendar_year,
    CASE WHEN EXTRACT(MONTH FROM rse.assessment_period) >= 9 OR EXTRACT(MONTH FROM rse.assessment_period) <= 1 THEN 'autumn' ELSE 'spring' END AS semester,
    
    -- Anomaly classification
    'MULTI_SOURCE_PATTERN' AS anomaly_type,
    
    -- From view_risk_score_evolution
    COUNT(DISTINCT rse.person_id) FILTER (WHERE rse.risk_severity IN ('HIGH', 'CRITICAL')) AS politician_count_with_risk,
    ROUND(AVG(rse.risk_score), 2) AS avg_risk_score,
    COUNT(*) FILTER (WHERE rse.severity_transition LIKE 'ESCALATION%') AS risk_escalations,
    
    -- From view_riksdagen_voting_anomaly_detection (NEW - uses actual columns: total_rebellions, anomaly_classification)
    COUNT(DISTINCT vad.person_id) FILTER (WHERE vad.anomaly_classification IN ('FREQUENT_STRONG_REBEL', 'CONSISTENT_REBEL')) AS high_anomaly_count,
    ROUND(AVG(vad.total_rebellions), 2) AS avg_total_rebellions,
    COUNT(DISTINCT vad.person_id) FILTER (WHERE vad.strong_consensus_rebellions >= 5) AS strong_consensus_rebels,
    
    -- From view_politician_risk_summary (NEW - aggregated risk profiles)
    ROUND(AVG(prs.risk_score), 2) AS avg_risk_score_prs,
    COUNT(DISTINCT prs.person_id) FILTER (WHERE prs.risk_level IN ('HIGH', 'CRITICAL')) AS high_risk_politicians
    
FROM election_cycle_periods ecp
LEFT JOIN view_risk_score_evolution rse
    ON EXTRACT(YEAR FROM rse.assessment_period) = ecp.calendar_year
LEFT JOIN view_riksdagen_voting_anomaly_detection vad
    ON 1=1  -- Current snapshot (3-year rolling window), no temporal filtering available
LEFT JOIN view_politician_risk_summary prs
    ON 1=1  -- Current snapshot, not time-filtered
GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, 
    CASE WHEN EXTRACT(MONTH FROM rse.assessment_period) >= 9 OR EXTRACT(MONTH FROM rse.assessment_period) <= 1 THEN 'autumn' ELSE 'spring' END
ORDER BY election_cycle_id, cycle_year, semester;
        ]]>
    </createView>
    <rollback><dropView viewName="view_election_cycle_anomaly_pattern"/></rollback>
</changeSet>


<!-- ========================================================================= -->
<!-- 4. PREDICTIVE INTELLIGENCE (ENHANCED): view_election_cycle_predictive_intelligence -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-predictive-004" failOnError="true">
    <comment>
        REVISED: Enhanced with view_ministry_risk_evolution and view_party_effectiveness_trends
        
        Framework: Predictive Intelligence (14 supporting views, 8/8 risk rules)
        
        Source Views (COMPREHENSIVE):
        - view_risk_score_evolution: Risk forecasting and trajectories
        - view_ministry_risk_evolution: Ministry-level risk trends (NEW)
        - view_party_effectiveness_trends: Party performance trajectories (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_predictive_intelligence">
        <![CDATA[
WITH election_cycle_periods AS (
    SELECT 
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) || '-' || (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4) AS election_cycle_id,
        (year_series - (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) + 1) AS cycle_year,
        year_series AS calendar_year
    FROM generate_series(1994, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
)
SELECT 
    ecp.election_cycle_id,
    ecp.cycle_year,
    ecp.calendar_year,
    CASE WHEN EXTRACT(MONTH FROM rse.assessment_period) >= 9 OR EXTRACT(MONTH FROM rse.assessment_period) <= 1 THEN 'autumn' ELSE 'spring' END AS semester,
    
    -- From view_risk_score_evolution (risk forecasting)
    CASE 
        WHEN AVG(rse.risk_score_change) >= 10 THEN 'RAPID_ESCALATION'
        WHEN AVG(rse.risk_score_change) >= 5 THEN 'MODERATE_ESCALATION'
        WHEN AVG(rse.risk_score_change) <= -5 THEN 'IMPROVING'
        ELSE 'STABLE'
    END AS risk_forecast_category,
    COUNT(DISTINCT rse.person_id) FILTER (WHERE rse.risk_trend LIKE '%INCREASE%') AS politicians_at_risk,
    ROUND(AVG(rse.risk_score_change), 2) AS avg_risk_score_change,
    
    -- From view_ministry_risk_evolution (NEW - government stability forecasting)
    COUNT(DISTINCT mre.name) FILTER (WHERE mre.risk_level IN ('HIGH', 'CRITICAL')) AS ministries_at_risk,
    ROUND(AVG(mre.rolling_avg_documents), 2) AS avg_ministry_productivity,
    
    -- From view_party_effectiveness_trends (NEW - party performance forecasting)
    ROUND(AVG(pet.win_rate_trend), 2) AS avg_party_win_rate_trend,
    COUNT(DISTINCT pet.party) FILTER (WHERE pet.absence_trend > 0) AS parties_with_increasing_absence
    
FROM election_cycle_periods ecp
LEFT JOIN view_risk_score_evolution rse
    ON EXTRACT(YEAR FROM rse.assessment_period) = ecp.calendar_year
LEFT JOIN view_ministry_risk_evolution mre
    ON EXTRACT(YEAR FROM mre.assessment_period) = ecp.calendar_year
LEFT JOIN view_party_effectiveness_trends pet
    ON EXTRACT(YEAR FROM pet.period_start) = ecp.calendar_year
GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, 
    CASE WHEN EXTRACT(MONTH FROM rse.assessment_period) >= 9 OR EXTRACT(MONTH FROM rse.assessment_period) <= 1 THEN 'autumn' ELSE 'spring' END
ORDER BY election_cycle_id, cycle_year, semester;
        ]]>
    </createView>
    <rollback><dropView viewName="view_election_cycle_predictive_intelligence"/></rollback>
</changeSet>

<!-- ========================================================================= -->
<!-- 5. NETWORK ANALYSIS (ENHANCED): view_election_cycle_network_analysis    -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-network-005" failOnError="true">
    <comment>
        REVISED: Enhanced with view_riksdagen_politician_influence_metrics
        
        Framework: Network Analysis (11 supporting views, 3/4 risk rules)
        
        Source Views (COMPREHENSIVE):
        - view_riksdagen_coalition_alignment_matrix: Coalition structure
        - view_riksdagen_politician_influence_metrics: Network centrality, power brokers (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_network_analysis">
        <![CDATA[
WITH election_cycle_periods AS (
    SELECT 
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) || '-' || (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4) AS election_cycle_id,
        (year_series - (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) + 1) AS cycle_year,
        year_series AS calendar_year
    FROM generate_series(2020, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
)
SELECT 
    ecp.election_cycle_id,
    ecp.cycle_year,
    ecp.calendar_year,
    'aggregate' AS semester,
    
    -- From view_riksdagen_coalition_alignment_matrix
    cam.party1,
    cam.party2,
    cam.alignment_rate,
    CASE 
        WHEN cam.alignment_rate >= 80 THEN 'STRONG_COALITION'
        WHEN cam.alignment_rate >= 60 THEN 'MODERATE_COALITION'
        ELSE 'WEAK_COALITION'
    END AS coalition_strength,
    
    -- From view_riksdagen_politician_influence_metrics (NEW - network analysis)
    COUNT(DISTINCT pim.person_id) FILTER (WHERE pim.influence_classification IN ('HIGH_INFLUENCE', 'VERY_HIGH_INFLUENCE')) AS influential_politicians,
    ROUND(AVG(pim.network_median), 2) AS avg_network_centrality,
    COUNT(DISTINCT pim.person_id) FILTER (WHERE pim.broker_classification = 'POWER_BROKER') AS power_broker_count
    
FROM election_cycle_periods ecp
CROSS JOIN view_riksdagen_coalition_alignment_matrix cam
LEFT JOIN view_riksdagen_politician_influence_metrics pim
    ON 1=1  -- Current snapshot
WHERE cam.party1 IS NOT NULL AND cam.party2 IS NOT NULL
GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, cam.party1, cam.party2, cam.alignment_rate
ORDER BY election_cycle_id, cycle_year, alignment_rate DESC;
        ]]>
    </createView>
    <rollback><dropView viewName="view_election_cycle_network_analysis"/></rollback>
</changeSet>


<!-- ========================================================================= -->
<!-- 6. DECISION INTELLIGENCE (ENHANCED): view_election_cycle_decision_intelligence -->
<!-- ========================================================================= -->

<changeSet author="intelligence-operative" id="1.51-election-cycle-decision-006" failOnError="true">
    <comment>
        REVISED: Enhanced with view_decision_temporal_trends and view_ministry_decision_impact
        
        Framework: Decision Intelligence (5 supporting views, 5/5 risk rules)
        
        Source Views (COMPREHENSIVE):
        - view_riksdagen_party_decision_flow: Party proposal outcomes
        - view_decision_temporal_trends: Temporal decision patterns (NEW)
        - view_ministry_decision_impact: Government decision effectiveness (NEW)
    </comment>
    
    <createView viewName="view_election_cycle_decision_intelligence">
        <![CDATA[
WITH election_cycle_periods AS (
    SELECT 
        (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) || '-' || (1994 + (FLOOR((year_series - 1994) / 4.0) * 4) + 4) AS election_cycle_id,
        (year_series - (1994 + (FLOOR((year_series - 1994) / 4.0) * 4)) + 1) AS cycle_year,
        year_series AS calendar_year
    FROM generate_series(1994, EXTRACT(YEAR FROM CURRENT_DATE)::INTEGER + 4, 1) AS year_series
)
SELECT 
    ecp.election_cycle_id,
    ecp.cycle_year,
    ecp.calendar_year,
    CASE WHEN pdf.decision_month_num >= 9 OR pdf.decision_month_num <= 1 THEN 'autumn' ELSE 'spring' END AS semester,
    pdf.party,
    
    -- From view_riksdagen_party_decision_flow (party-level outcomes)
    SUM(pdf.total_proposals) AS total_proposals,
    SUM(pdf.approved_proposals) AS approved_proposals,
    SUM(pdf.rejected_proposals) AS rejected_proposals,
    ROUND(AVG(pdf.approval_rate), 2) AS avg_approval_rate,
    
    -- Decision effectiveness classification
    CASE 
        WHEN AVG(pdf.approval_rate) >= 75 THEN 'HIGHLY_EFFECTIVE'
        WHEN AVG(pdf.approval_rate) >= 50 THEN 'MODERATELY_EFFECTIVE'
        ELSE 'LOWLY_EFFECTIVE'
    END AS decision_effectiveness,
    
    -- From view_decision_temporal_trends (NEW - temporal decision patterns)
    ROUND(AVG(dtt.daily_approval_rate), 2) AS temporal_approval_rate,
    SUM(dtt.daily_decisions) AS temporal_decision_count,
    
    -- From view_ministry_decision_impact (NEW - government decision effectiveness)
    ROUND(AVG(mdi.approval_rate), 2) AS ministry_impact_score,
    COUNT(DISTINCT mdi.ministry_code) AS ministries_with_decisions
    
FROM election_cycle_periods ecp
LEFT JOIN view_riksdagen_party_decision_flow pdf
    ON pdf.decision_year = ecp.calendar_year
LEFT JOIN view_decision_temporal_trends dtt
    ON EXTRACT(YEAR FROM dtt.decision_day) = ecp.calendar_year
LEFT JOIN view_ministry_decision_impact mdi
    ON mdi.decision_year = ecp.calendar_year
WHERE pdf.party IS NOT NULL
GROUP BY ecp.election_cycle_id, ecp.cycle_year, ecp.calendar_year, 
    CASE WHEN pdf.decision_month_num >= 9 OR pdf.decision_month_num <= 1 THEN 'autumn' ELSE 'spring' END, 
    pdf.party
HAVING SUM(pdf.total_proposals) > 0
ORDER BY election_cycle_id, cycle_year, semester, party;
        ]]>
    </createView>
    <rollback><dropView viewName="view_election_cycle_decision_intelligence"/></rollback>
</changeSet>

<!-- ========================================================================= -->
<!-- VALIDATION: Verify Election Cycle Views Created Successfully           -->
<!-- ========================================================================= -->

<changeSet id="1.51-validation" author="intelligence-operative">
    <comment>
        Validate that all 6 election cycle views were created successfully
        with enhanced comprehensive META/META-level integration.
    </comment>
    
    <sql splitStatements="false"><![CDATA[
        DO $$
        DECLARE
            v_temporal_exists BOOLEAN;
            v_comparative_exists BOOLEAN;
            v_pattern_exists BOOLEAN;
            v_predictive_exists BOOLEAN;
            v_network_exists BOOLEAN;
            v_decision_exists BOOLEAN;
        BEGIN
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_temporal_trends'
            ) INTO v_temporal_exists;
            
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_comparative_analysis'
            ) INTO v_comparative_exists;
            
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_anomaly_pattern'
            ) INTO v_pattern_exists;
            
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_predictive_intelligence'
            ) INTO v_predictive_exists;
            
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_network_analysis'
            ) INTO v_network_exists;
            
            SELECT EXISTS(
                SELECT 1 FROM information_schema.views 
                WHERE table_name = 'view_election_cycle_decision_intelligence'
            ) INTO v_decision_exists;
            
            IF v_temporal_exists AND v_comparative_exists AND v_pattern_exists 
               AND v_predictive_exists AND v_network_exists AND v_decision_exists THEN
                RAISE NOTICE '✓ All 6 ENHANCED election cycle views created successfully';
                RAISE NOTICE '  REVISION: Now using comprehensive advanced view integration';
                RAISE NOTICE '  - Temporal: Added decision_temporal_trends, committee_productivity';
                RAISE NOTICE '  - Comparative: Added party_performance_metrics, committee_productivity_matrix';
                RAISE NOTICE '  - Pattern: Added voting_anomaly_detection, politician_risk_summary';
                RAISE NOTICE '  - Predictive: Added ministry_risk_evolution, party_effectiveness_trends';
                RAISE NOTICE '  - Network: Added politician_influence_metrics';
                RAISE NOTICE '  - Decision: Added decision_temporal_trends, ministry_decision_impact';
            ELSE
                RAISE WARNING 'Some election cycle views failed to create';
            END IF;
        END $$;
    ]]></sql>
</changeSet>

</databaseChangeLog>