Admission Advisor

Analytical platform for Indonesian university Jalur Mandiri admissions. Identifies Sarjana (S1) programs with highest acceptance probability at minimum passing scores across 5 top national universities.

Live: uni-advisor-kappa.vercel.app

Built with an autonomous scraping pipeline that attempts official portals first, falls back to affiliated platforms, and defaults to curated public estimates. All data is transparent — every figure traces back to its source.

---

Features

• Real-time scraping — Playwright + Cheerio pipeline targeting official university admissions portals
• Multi-source fallback — Official → Affiliated (Zenius/Quipper/SNPMB) → Curated public estimates
• Ease Score ranking — Composite metric favouring programs with highest acceptance probability at lowest passing score
• Volatility Index — Measures historical instability (coefficient of variation) of a program's minimum passing score across 3 years
• Crowding Penalty — Penalises programs with sharp score drops that attract applicant surges the following cycle (mean-reversion trap detection)
• Data Archives — Full dataset transparency with per-program source citations and verified quota figures
• Score trend analysis — 3-year minimum score trends (2022–2024) with directional indicators
• Mobile responsive — Optimised for phones (390px+), tablets, and desktop viewports

Supported Universities

ID	University	Mandiri Portal
ugm	Universitas Gadjah Mada	UM-UGM
ui	Universitas Indonesia	SIMAK UI
itb	Institut Teknologi Bandung	USM ITB
its	Institut Teknologi Sepuluh Nopember	SM ITS
unpad	Universitas Padjadjaran	SMUP Unpad

---

Quick Start

# Install dependencies
npm install

# Install Playwright Chromium (required for JS-rendered pages)
npx playwright install chromium

# Start the server
npm start

Open http://localhost:3000

---

Deployment

Hosted on Vercel with automatic deploys from master:

npx vercel --prod

---

API

Method	Endpoint	Description
GET	/api/universities	List supported universities
GET	/api/admissions/:id	Admissions data with recommendations (cache → scrape → fallback)
GET	/api/admissions/:id/refresh	Force fresh scrape, bypass cache
GET	/api/archives/:id	Raw dataset with full source citations
GET	/api/health	Server health and cache status

Response Fields (Admissions)

Each program object includes:

Field	Description
ease_score	Composite ranking metric (higher = easier to enter)
volatility_index	Coefficient of variation of 3-year min scores (σ/μ)
sharp_drop_pct	Year-over-year score drop percentage (0 if ≤3%)
crowding_penalty	(volatility × 3.0) + (sharpDrop × 5.0), capped at 2.0
crowding_risk	low / moderate / high based on penalty thresholds

---

Project Structure

uni-advisor/
├── src/
│   ├── scrapers/           # Per-university scraper modules
│   │   ├── index.js        # Scraper dispatcher (official → affiliated → fallback)
│   │   ├── ugm.js
│   │   ├── ui.js
│   │   ├── itb.js
│   │   ├── its.js
│   │   └── unpad.js
│   ├── strategies/
│   │   ├── official.js     # Playwright + Cheerio scraping
│   │   ├── affiliated.js   # Third-party platform scrapers
│   │   └── fallback.js     # Curated JSON loader with source passthrough
│   ├── utils/
│   │   ├── parser.js       # Raw → unified schema normaliser
│   │   ├── scorer.js       # Ease Score + Volatility Index + Crowding Penalty
│   │   └── cache.js        # File-based JSON cache (24h TTL)
│   └── server.js           # Express API
├── public/
│   └── index.html          # Single-file frontend (HTML + CSS + JS)
├── data/
│   ├── cache/              # Auto-generated per-university cache
│   └── hardcoded/          # Curated fallback datasets with source citations
│       ├── ugm.json        # 36 programs, 4 sources
│       ├── ui.json         # 26 programs, 4 sources
│       ├── itb.json        # 26 programs, 3 sources
│       ├── its.json        # 25 programs, 3 sources
│       └── unpad.json      # 26 programs, 3 sources
├── directives/
│   └── system_context.md   # Full operational context for agents
├── vercel.json             # Vercel deployment config
├── .env
├── package.json
└── README.md

---

Data Sources & Transparency

Each program entry in the dataset includes source_ref indices pointing to verified citation objects. The Data Archives section in the UI makes these fully browsable.

Source types:

Type	Reliability	Description
official-portal	High	Direct from university admissions portal
news-report	Medium–High	Published by Kontan, Tempo, Detik, Sindonews, etc.
education-platform	Medium	Aggregated by UTBK-CAK, Brain Academy, Zenius, etc.

Citation schema:

{
  "id": 0,
  "title": "Daya Tampung Jalur Mandiri UGM 2024",
  "publisher": "Kontan",
  "url": "https://...",
  "type": "news-report",
  "accessed": "2024-06-15",
  "reliability": "high",
  "notes": "Per-prodi breakdown: Teknik Sipil 52, Hukum 128, Kedokteran 69."
}

---

Ease Score Formula

Programs are ranked by a composite Ease Score:

ease_score = (acceptance_rate × 0.6) + ((1 / avg_min_score) × 10000 × 0.4) + trend_bonus − crowding_penalty

Component	Weight	Logic
Acceptance rate	60%	Higher rate = easier admission
Inverse score	40%	Lower min score = more accessible
Trend bonus	±0.5	Decreasing = +0.5, Increasing = -0.5
Crowding penalty	0–2.0	Penalises volatile programs with sharp drops

Crowding Penalty

crowding_penalty = min((volatility_index × 3.0) + (sharp_drop × 5.0), 2.0)

Risk Level	Penalty Range	Meaning
Low	< 0.3	Stable, safe to recommend
Moderate	0.3 – 1.0	Some volatility, monitor closely
High	≥ 1.0	Likely applicant surge, avoid recommending

---

Environment Variables

Variable	Default	Description
PORT	3000	HTTP server port
PLAYWRIGHT_HEADLESS	true	Run Chromium headlessly
CACHE_TTL_HOURS	24	Cache validity (hours)
NODE_ENV	development	Environment mode

---

Adding a University

1. Create src/scrapers/<id>.js exporting officialUrls, affiliatedUrls, scrapeOfficial(), scrapeAffiliated()
2. Register in src/scrapers/index.js
3. Add to UNIVERSITIES array in src/server.js
4. Create data/hardcoded/<id>.json with programs and sources citation array
5. Add affiliated URLs in src/strategies/affiliated.js

---

License

MIT

---

Disclaimer

Data sourced from official university admissions portals and affiliated educational platforms. Historical minimum passing scores are estimates derived from publicly available records. This tool is for informational and educational purposes only and is not affiliated with any university or government body.