A data analysis of Timor-Leste's occupation and post-conflict recovery, 1990–2012, integrating three independent real-world data sources into a single year-indexed dataset and testing whether the 1999 independence referendum and the UN interventions that followed show up in the numbers.
- State violence dropped sharply after 1999. The Political Terror Scale score averaged 4.00 during the occupation (1990–1999) and 2.23 during recovery (2000–2012) — a fitted decline of 0.120 points per year.
- Displacement peaked at 162,472 registered refugees in 1999, then collapsed to a few hundred within four years as return waves ran through 2002.
- Arms transfers and political terror are essentially uncorrelated here (r = −0.018). That cuts against the common assumption that arms supply tracks violence, and it argues for treating arms flows and on-the-ground violence as separate indicators rather than proxies for each other.
- Arms transfers to Indonesia trended upward across the whole period (+3.91 TIV/yr), peaking at 578 TIV in 2007 — two years after the U.S. embargo was lifted, and well into Timor-Leste's recovery.
Full write-up with citations: REPORT.md. Why each source was chosen, and its limitations: SOURCES.md.
| Source | Used for | File |
|---|---|---|
| UNHCR Refugee Data Finder | Refugees originating from Timor-Leste, by year | data/un_unhcr_refugees.csv |
| SIPRI Arms Transfers Database | Arms transfers to Indonesia (Trend Indicator Values) | data/sipri_arms_indonesia.csv |
| Political Terror Scale | State-inflicted political violence, 1–5 scale | data/political_terror_scale.csv |
The CSVs here are extracts pulled from those databases for this analysis. They remain the work of their respective publishers — cite the original source, not this repo, if you reuse them.
Because East Timor was not recognized as a separate country of origin during the occupation, UNHCR records no displacement from it before 1999. PTS scores are therefore read from Indonesia for 1990–1999 (the occupying power) and Timor-Leste for 2000–2012, which is what makes the before/after comparison possible at all.
pip install -r requirements.txt
python analysis.pyWrites east_timor_analysis.png (4-panel figure) and data_summary.csv (key findings) to the
project root. analysis.ipynb is the same analysis as a notebook, broken into cells.
This began as the capstone for an introductory data engineering course. The assignment was to take one real conflict or post-conflict situation, pull a required UN dataset plus two additional sources of your own choosing, justify those choices in writing, integrate everything into one dataset, and answer three research questions with supporting charts and a short report. I chose Timor-Leste and the post-conflict recovery track, and answered the pattern comparison, trend analysis, and policy effectiveness questions.
Code released under the MIT License. The report text and the underlying datasets are not covered by it — see the data sources table above.