Skip to content

test.md

Latest commit

 

History

History
309 lines (225 loc) · 9.87 KB

test.md

File metadata and controls

309 lines (225 loc) · 9.87 KB

PJT Human Capital Mapping System - Process Flowchart

System Overview

flowchart TB
    Start([User Opens Application]) --> LoadData[Load Data from Database]
    LoadData --> DataCache[(Data Cache:<br/>Persons, Deals, Companies,<br/>Roles, History)]
    
    DataCache --> Tab1[Top Affiliated Persons]
    DataCache --> Tab2[Top Deal Makers]
    DataCache --> Tab3[Connection Pathfinder]
    
    style DataCache fill:#4fc3f7
    style Tab1 fill:#ffb74d
    style Tab2 fill:#81c784
    style Tab3 fill:#ba68c8
Loading

1. Top Affiliated Persons Tab

Purpose: Rank persons by their total affiliations (companies worked at)

flowchart TD
    Start1([Top Affiliated Persons]) --> GetPersons[Get All Persons<br/>from Database]
    GetPersons --> CountAff[Count Affiliations<br/>per Person]
    
    CountAff --> Query{User Filters?}
    Query -->|Min Affiliations| Filter1[Filter by<br/>Min Count]
    Query -->|Limit| Filter2[Limit Results]
    
    Filter1 --> Sort[Sort by<br/>Affiliation Count DESC]
    Filter2 --> Sort
    
    Sort --> Display1[Display Table:<br/>- Name<br/>- Affiliation Count<br/>- Current Company<br/>- Current Position]
    
    Display1 --> Export1[Export to CSV]
    
    style Start1 fill:#ffb74d
    style Display1 fill:#2e7d32,color:#fff
Loading

Key Metrics:

  • Affiliation Count: Total number of companies person has worked at (from df_history)
  • Sorting: Descending by affiliation count
  • Filters: Minimum affiliations threshold, result limit

2. Top Deal Makers Tab

Purpose: Rank persons by their involvement in PJT-advised deals

flowchart TD
    Start2([Top Deal Makers]) --> BuildGraph[Build Connection Graph]
    
    BuildGraph --> Step1[1. Find PJT Advisors<br/>on ALL Deals]
    Step1 --> Step2[2. Match Persons<br/>to Deal Companies]
    Step2 --> Validate{Person Active<br/>During Deal?}
    Validate -->|Yes| AddPerson[Add to Deal Persons]
    Validate -->|No| Skip[Skip Person]
    
    AddPerson --> Aggregate[Aggregate per Person:<br/>- Deal Count<br/>- Total Deal Value<br/>- Company Count]
    
    Aggregate --> Sort2[Sort by<br/>Deal Count DESC]
    Sort2 --> Display2[Display Table:<br/>- Name<br/>- # Deals<br/>- Total Deal Value<br/>- # Companies<br/>- Current Role<br/>- Action Button]
    
    Display2 --> Action{User Action?}
    Action -->|Click Find Connection| GoToTab3[Navigate to<br/>Connection Pathfinder]
    Action -->|Export| Export2[Export to CSV]
    
    style Start2 fill:#81c784
    style Display2 fill:#2e7d32,color:#fff
    style GoToTab3 fill:#ba68c8
Loading

Key Logic:

  • No Date Filter: Includes ALL PJT deals (no 2015 filter)
  • Deal Matching: Person must be active at target/acquiror company during deal announcement
  • Active Check: position_start_date <= deal_date <= position_end_date
  • Metrics: Count of deals, sum of deal values, count of unique companies

3. Connection Pathfinder Tab

Purpose: Find shortest valid connection path from PJT Partners to target person

3.1 Graph Building Process

flowchart TD
    Start3([Connection Pathfinder]) --> Input[User Enters<br/>Person ID]
    Input --> CheckCache{Base Graph<br/>Cached?}
    
    CheckCache -->|No| BuildBase[Build Base Graph]
    CheckCache -->|Yes| UseCache[Use Cached Graph]
    
    BuildBase --> FilterDeals[Filter Deals<br/>After 2015-01-01]
    FilterDeals --> AddNodes[Add Graph Nodes]
    
    AddNodes --> Node1[ADVISOR Nodes:<br/>PJT Advisors]
    AddNodes --> Node2[DEAL Nodes:<br/>PJT-Advised Deals]
    AddNodes --> Node3[PERSON Nodes:<br/>Deal-Related Persons]
    AddNodes --> Node4[COMPANY Nodes:<br/>Shared Companies]
    
    Node1 --> AddEdges[Add Graph Edges]
    Node2 --> AddEdges
    Node3 --> AddEdges
    Node4 --> AddEdges
    
    AddEdges --> Edge1[ADVISOR→DEAL:<br/>PJT advised on deal]
    AddEdges --> Edge2[DEAL→PERSON:<br/>Person active at<br/>deal company]
    AddEdges --> Edge3[PERSON→COMPANY:<br/>Overlapping colleagues]
    
    Edge1 --> CacheBase[(Cache Base Graph)]
    Edge2 --> CacheBase
    Edge3 --> CacheBase
    
    CacheBase --> UseCache
    
    style Start3 fill:#ba68c8
    style CacheBase fill:#4fc3f7
Loading

3.2 Graph Expansion & Pathfinding

flowchart TD
    BaseGraph[(Base Graph:<br/>178 persons)] --> CheckTarget{Target Person<br/>in Base Graph?}
    
    CheckTarget -->|Yes| UseDirect[Use Base Graph]
    CheckTarget -->|No| Expand[Expand Graph]
    
    Expand --> Pass1[Pass 1:<br/>Find All Companies<br/>of Base Persons]
    Pass1 --> Pass2[Pass 2:<br/>Find All Persons<br/>at Those Companies<br/>+ Target Person]
    
    Pass2 --> BuildEdges[Build PERSON-COMPANY<br/>Edges with Overlap Check]
    
    BuildEdges --> OverlapCheck{Persons Overlapped<br/>at Company?}
    OverlapCheck -->|Yes| AddEdge[Add Edge with<br/>Time Periods]
    OverlapCheck -->|No| SkipEdge[Skip Edge]
    
    AddEdge --> CacheExpanded[(Cache Expanded Graph<br/>for This Person)]
    SkipEdge --> CacheExpanded
    
    CacheExpanded --> UseDirect
    UseDirect --> FilterDeals2[Filter to Deals<br/>After 2015-01-01]
    FilterDeals2 --> FindPaths[Find All Paths<br/>Using NetworkX]
    
    style BaseGraph fill:#4fc3f7
    style CacheExpanded fill:#4fc3f7
Loading

3.3 Path Validation & Selection

flowchart TD
    AllPaths[All Potential Paths<br/>from NetworkX] --> Loop{For Each Path}
    
    Loop --> ValidateDeal[Validate Deal Connection]
    
    ValidateDeal --> CheckDirector{First Director<br/>Active During<br/>Deal Date?}
    CheckDirector -->|No| Reject1[❌ Reject Path:<br/>Director Not Active]
    CheckDirector -->|Yes| ValidateOverlaps[Validate All<br/>Person Overlaps]
    
    ValidateOverlaps --> LoopConnections{For Each<br/>PERSON→COMPANY→PERSON}
    
    LoopConnections --> CheckOverlap{Time Periods<br/>Overlap?}
    CheckOverlap -->|No| Reject2[❌ Reject Path:<br/>No Overlap]
    CheckOverlap -->|Yes| NextConnection[Check Next<br/>Connection]
    
    NextConnection --> MoreConnections{More<br/>Connections?}
    MoreConnections -->|Yes| LoopConnections
    MoreConnections -->|No| Accept[✅ Accept Path:<br/>All Valid]
    
    Accept --> ValidPaths[Valid Paths List]
    Reject1 --> Loop
    Reject2 --> Loop
    
    ValidPaths --> SelectShortest[Select Shortest<br/>Valid Path]
    SelectShortest --> DisplayPath[Display Path<br/>in UI]
    
    DisplayPath --> ShowSteps[Show Each Step:<br/>- Node Name<br/>- Connection Detail<br/>- Time Periods<br/>- Deal Dates]
    
    style Accept fill:#2e7d32,color:#fff
    style Reject1 fill:#c62828,color:#fff
    style Reject2 fill:#c62828,color:#fff
    style DisplayPath fill:#2e7d32,color:#fff
Loading

3.4 Validation Rules Detail

flowchart LR
    subgraph "Deal Validation"
        Deal[DEAL Node] --> Person1[PERSON Node]
        Person1 --> Check1{Person Active<br/>at Deal Company<br/>During Deal Date?}
        Check1 -->|Yes| Valid1[✅ Valid]
        Check1 -->|No| Invalid1[❌ Invalid]
    end
    
    subgraph "Overlap Validation"
        Person2[PERSON A] --> Company[COMPANY Node]
        Company --> Person3[PERSON B]
        
        Person2 -.->|Works: 2010-2015| TimeA[Time Period A]
        Person3 -.->|Works: 2012-2020| TimeB[Time Period B]
        
        TimeA --> Check2{Periods Overlap?<br/>2012-2015}
        TimeB --> Check2
        Check2 -->|Yes| Valid2[✅ Valid]
        Check2 -->|No| Invalid2[❌ Invalid]
    end
    
    style Valid1 fill:#2e7d32,color:#fff
    style Invalid1 fill:#c62828,color:#fff
    style Valid2 fill:#2e7d32,color:#fff
    style Invalid2 fill:#c62828,color:#fff
Loading

Data Flow Between Components

flowchart LR
    subgraph Database
        DB[(MySQL Database)]
    end
    
    subgraph Backend
        Cache[(Data Cache)]
        Graph[(Connection Graph)]
    end
    
    subgraph Frontend
        Tab1[Top Affiliated<br/>Persons]
        Tab2[Top Deal<br/>Makers]
        Tab3[Connection<br/>Pathfinder]
    end
    
    DB -->|Load on Startup| Cache
    Cache -->|Build Graph| Graph
    
    Cache -->|Count Affiliations| Tab1
    Graph -->|Count Deals| Tab2
    Graph -->|Find Path| Tab3
    
    Tab2 -.->|Click Find Connection| Tab3
    
    style DB fill:#ef6c00
    style Cache fill:#4fc3f7
    style Graph fill:#4fc3f7
Loading

Key Performance Optimizations

flowchart TD
    Opt1[Caching Strategy] --> Cache1[Base Graph Cache:<br/>Shared Across Searches]
    Opt1 --> Cache2[Expanded Graph Cache:<br/>Per-Person Cache]
    
    Opt2[Graph Building] --> Build1[Filter Deals After 2015<br/>(Pathfinder Only)]
    Opt2 --> Build2[Only Overlap Check<br/>for Edges]
    Opt2 --> Build3[Skip Companies with<br/>>50 Persons]
    
    Opt3[Validation] --> Val1[Early Rejection:<br/>Invalid Deal Director]
    Opt3 --> Val2[Early Rejection:<br/>No Overlap]
    
    style Opt1 fill:#1565c0,color:#fff
    style Opt2 fill:#1565c0,color:#fff
    style Opt3 fill:#1565c0,color:#fff
Loading

Summary of Key Algorithms

Top Affiliated Persons

  • Algorithm: Simple aggregation and sort
  • Complexity: O(n) where n = number of persons
  • Data Source: df_history table

Top Deal Makers

  • Algorithm: Graph-based deal matching with temporal validation
  • Complexity: O(d × p) where d = deals, p = persons per deal
  • Data Source: df_deals, df_history, df_fees
  • Note: Includes ALL PJT deals (no date filter)

Connection Pathfinder

  • Algorithm:
    1. Filter deals to after 2015-01-01
    2. NetworkX shortest path (BFS)
    3. Post-processing validation (temporal overlap)
  • Complexity: O(V + E) for pathfinding + O(p²) for validation where p = path length
  • Data Source: All tables combined into graph structure
  • Note: Only includes deals announced after 2015-01-01

Validation Rules

  1. Deal Connection: Director must be active at deal company during deal announcement
  2. Person Connection: Two persons must have overlapping work periods at the same company
  3. Time Overlap: start1 <= end2 AND start2 <= end1