Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/HelenDiMo/archaios-data-Intelligence/llms.txt

Use this file to discover all available pages before exploring further.

The A.D.I. platform operates on a two-layer data architecture. The foundation is five raw CSV files in the /data/ directory, sourced directly from the U.S. Department of Defense THOR (Theater History of Operations Reports) database — these files are never modified and serve as the read-only historical record. On top of that foundation, A.D.I.’s Jupyter notebook EDA pipeline produces three enriched ADI master CSVs at the repository root, each scoped to a specific chronological phase of the Normandy campaign. These master files are the tables that Power BI ingests, forming a semantic model (desembarco_normandia_dashboard.SemanticModel) that drives the interactive dashboard used by archaeological prospection teams.

​
Architecture Diagram

THOR Source Data (data/)
  1_cleanbombww2.csv
  2_operations.csv          →  adi_master_1943_1944_CLEAN.csv  ─┐
  3_thor_wwii_aircraft_gloss.csv                                  ├→  Power BI Semantic Model
  4_thor_wwii_data_clean.csv   →  adi_master_1_5_junio.csv  ────┤    (3 tables + relationships)
  5_thor_wwii_weapon_gloss.csv  →  adi_dataset_master.csv  ─────┘

​
Data Layers

​
THOR Source Layer

The five files in the /data/ directory are raw and cleaned exports from the DoD THOR database, a declassified archive of Allied aerial bombing records for World War II. They are read-only reference data — no rows are added or modified by the A.D.I. pipeline. They provide the unaltered historical record from which the ADI master datasets are derived.
FileContent
1_cleanbombww2.csvAerial incursion records — primarily MTO theater
2_operations.csvFull THOR operations dataset — all theaters and missions
3_thor_wwii_aircraft_gloss.csvGlossary decoding Aircraft Series codes to model names
4_thor_wwii_data_clean.csvCleaned full THOR WWII dataset covering all theaters
5_thor_wwii_weapon_gloss.csvGlossary decoding ordnance codes to descriptive names

​
ADI Master Layer

The three enriched CSV files at the repository root are the output of A.D.I.’s data cleaning and feature-engineering pipeline (desembarco_normandia.ipynb). Each file is scoped to a distinct chronological phase of the Normandy campaign and contains a curated, cleaned subset of THOR fields plus A.D.I.’s own derived columns (notably weapon_type). These are the files that Power BI loads at startup.
FilePhaseDashboard Page
adi_master_1943_1944_CLEAN.csvAug 1943 – Jun 5, 1944 (pre-D-Day campaign)“Panorama 1943–1944”
adi_master_1_5_junio.csvJune 1–5, 1944 (logistical softening phase)“1–5 de Junio”
adi_dataset_master.csvJune 6, 1944 (D-Day)“Día D”

​
Semantic Model

The Power BI Fabric semantic model (desembarco_normandia_dashboard.SemanticModel) is configured with Spanish locale (es-ES) and Power BI V3 data source version. It imports the three ADI master CSVs in import mode, meaning data is cached in the model rather than queried live. Three analytical tables form the core of the model, in query order as defined in model.tmdl:
  1. adi_dataset_master — D-Day mission records (49 columns including A.D.I. enrichment fields)
  2. adi_master_1_5_junio — June 1–5, 1944 pre-invasion records (46 columns)
  3. adi_master_1943_1944 — August 1943 to June 5, 1944 campaign records (46 columns)
Time intelligence is enabled (__PBI_TimeIntelligenceEnabled = 1). Each table’s Mission Date column is linked to a dedicated local date table (LocalDateTable_*) via a datePartOnly join, enabling Power BI’s built-in date hierarchy — labeled as “Jerarquía de fechas” in the Spanish locale UI. Cross-filter relationships connect the three analytical tables through their shared Mission ID key, enabling slicers and filters applied to one table to propagate across the model.

​
Table Relationships

The relationships.tmdl file defines five relationships — three for time intelligence, two for cross-table filtering:

​
Cross-Table Relationships

From TableColumn→ To TableColumnCardinalityFilter Direction
adi_dataset_masterMission ID→ adi_master_1943_1944Mission IDOne-to-manyBidirectional
adi_master_1_5_junioMission ID→ adi_master_1943_1944Mission IDOne-to-manyBidirectional
Both cross-table relationships were auto-detected by Power BI (AutoDetected_* IDs) and use crossFilteringBehavior: bothDirections with fromCardinality: one, making each a one-to-many relationship. A filter on any one table’s Mission ID propagates to both other tables, enabling unified slicers across the three dashboard pages.

​
Date Table Relationships

Each analytical table has its own local date table linked by Mission Date using datePartOnly join behavior:
Analytical TableLocal Date Table
adi_dataset_masterLocalDateTable_42f1978c-4309-4622-a6e6-44f638248ae0
adi_master_1_5_junioLocalDateTable_5d16dce2-cfe9-4861-80dd-c3910e031179
adi_master_1943_1944LocalDateTable_16518298-1804-42b8-9956-d9f0d8219f62
The datePartOnly join means Power BI matches on the date portion of the Mission Date datetime column, ignoring any time component.

​
Next Steps

Dataset Reference

Detailed descriptions of all eight CSV files — five THOR source files and three ADI master datasets, with column lists and data quality notes.

Schema Reference

Complete field-by-field reference for the adi_dataset_master table — every column, its Power BI data type, and archaeological significance.

Build docs developers (and LLMs) love

Get started for freeTalk to us