muumaaz/spbu-applicants-2026

🎓 SPbU Foreign Applicants 2026 — Dataset & Analysis

Saint Petersburg State University (СПбГУ) — International applicant data for the 2026 academic year.

Metric	Value
Total application rows	3,680
Unique applicants	3,426
Countries represented	113
Budget-funded placements	784 (21.3%)
Degree types	Bachelor, Master, Specialist, PhD

---

📁 Repository Contents

File	Description
spbu_applicants_2026.csv	Main dataset (6 columns, 3,680 rows)
scrape_spbu.py	Scraper script that generated the CSV
analysis.ipynb	Complete Jupyter notebook with analysis, visualizations, and statistical tests
images/	All 17 visualization PNGs (150 DPI)

---

📊 Dataset Schema

Column	Type	Description
UID	int64	Unique applicant registration number (format: 26XXXXXX)
Country	string	Country of citizenship (Russian, with trailing /)
Programs_Applied_To	string	Semicolon-separated list of programs
Count_of_Programs	int64	Number of programs applied to (1–3)
Status	string	Accepted, Accepted with budget, or Not Accepted
Applicant	string	Degree type: Bachelor, Master, Specialist, or PhD

Note: 254 UIDs appear twice (applying to both Bachelor AND Specialist programs), bringing the total to 3,680 rows from 3,426 unique individuals.

---

🔍 Data Quality Report

Missing Values

Zero missing values across all 6 columns and all 3,680 rows. No blank strings, no NaN values. The scraper creates a row only when all fields are successfully extracted.

Outlier Analysis

Count_of_Programs is the only numeric variable. It is system-bounded to [1, 3] by the SPbU application portal:

Programs	Count	Percentage
1	1,726	46.9%
2	926	25.2%
3	1,028	27.9%

• IQR method: Q1=1, Q3=3, IQR=2 → fences at [-2, 6] → 0 outliers
• Z-score method: Max |z| = 1.41 → 0 outliers (threshold: 3)
• Rationale: No outlier removal needed — all values are valid, system-enforced bounds

Consistency

• ✅ All multi-row UIDs have identical Country values
• ✅ All multi-row UIDs have identical Status values
• ✅ Count_of_Programs matches actual semicolon count in 100% of rows

Data Cleaning Applied

Step	Rationale
Country name normalization	Stripped trailing /, normalized КИТАЙ → Китай, expanded abbreviations (ИРАН, ИСЛАМСКАЯ РЕСПУБЛИКА → Иран)
English translations	Added for visualization readability
Region mapping	Grouped 113 countries into 14 world regions for macro-level analysis
Program language extraction	Parsed (in English) / (in Russian) from program names
CIS flag	Identified former Soviet states for comparative analysis

---

📈 Key Visualizations

Overview Dashboard

Status Distribution

Interpretation: 78.5% of applications are accepted (self-funded), 21.3% receive budget (government) funding, and just 0.2% (8 rows) were not accepted. Budget status is determined by olympiad performance — winners with green-highlighted rows in official PDF results receive government funding.

Applicant Type Distribution

Interpretation: Bachelor's programs dominate (43.2%), followed by Master's and Specialist at near-parity (~24% each). PhD applications constitute 9.4%. The Bachelor dominance reflects SPbU's strong undergraduate recruitment pipeline from CIS countries.

Top 20 Countries (Histogram)

Interpretation: Kazakhstan leads with 588 applications (16%). The top 5 countries (Kazakhstan, Nigeria, Pakistan, Uzbekistan, China) account for 48.4% of all applications. The diversity spans Central Asia, Sub-Saharan Africa, South Asia, and the Middle East.

Country × Applicant Type Breakdown

Interpretation: CIS countries skew heavily toward Bachelor programs. South Asian countries (Pakistan, Bangladesh) show more Master/PhD interest. Iran is almost exclusively Specialist (medical degrees).

Budget Acceptance Rate by Country

Interpretation: Budget rates range from 71.4% (Kyrgyzstan) to near 0% for most non-CIS countries. The red dashed line marks the 21.3% overall rate. All countries above this line are CIS member states — reflecting Russia's bilateral education funding agreements.

Budget Rate by Degree Type

Interpretation: Bachelor programs have the highest budget rate (28.6%), nearly double Master's (16.4%). Specialist programs are lowest (11.5%). Budget scholarships for international students are disproportionately allocated at the undergraduate level.

Programs Distribution

Interpretation: PhD applicants are highly focused (74.8% apply to just 1 program), while Bachelor applicants are exploratory (mean 2.06 programs). This reflects increasing specialization at higher degree levels.

Regional Analysis

Interpretation: Central Asia dominates both volume (28.7%) and budget rate (~48%). Western Europe has almost no representation (8 applications total). SPbU primarily attracts applicants from developing countries and former Soviet states.

Budget Heatmap: Country × Degree Type

Interpretation: Kazakhstan shows high budget rates across Bachelor (50%) and Master (45%) but lower for Specialist (13%). Most non-CIS countries show 0% budget rate across all degree types, highlighting the sharp CIS advantage.

Program Language Preference

Interpretation: 58.3% of applications target Russian-medium programs. CIS applicants overwhelmingly choose Russian; English programs attract the most diverse international cohort.

Multi-Degree Applicants

Interpretation: 254 applicants (7.4%) apply to multiple degree types — 253 of these are Bachelor + Specialist combinations. This is a rational hedging strategy between 4-year and 5-year programs.

Status × Applicant Type

Top Programs

Interpretation: Management (in English) is the most popular program (434 applications), followed by Medicine (in Russian) and International Relations. Medical programs collectively dominate the Specialist category.

Programs Count vs Budget

Geographic Concentration (Lorenz Curve)

Interpretation: Gini coefficient of 0.799 indicates extreme concentration. The bottom 50% of countries contribute <5% of applications. International student mobility is dominated by a few high-volume corridors.

CIS vs Non-CIS Comparison

Interpretation: The CIS vs Non-CIS divide is the dataset's defining characteristic. CIS applicants have a 49.4% budget rate vs 7.1% for non-CIS — an odds ratio of 12.7×.

---

📊 Statistical Analysis Summary

All tests use α = 0.05 significance level.

#	Test	Variables	Result	Effect Size	Interpretation
1	Chi-squared	Status × Applicant Type	χ²=190.74, p<10⁻³⁸	Cramér's V=0.161	Budget rates differ significantly across degree types
2	Chi-squared	Budget × CIS	χ²=870.78, p<10⁻¹⁹¹	OR=12.71	CIS applicants are 12.7× more likely to get budget funding
3	Kruskal-Wallis	Programs Count × Degree Type	H=153.77, p<10⁻³³	—	PhD applicants apply to fewer programs than Bachelor
4	Mann-Whitney U	Programs Count: Budget vs Not	U=1,220,682, p=0.0005	Small	Budget applicants apply to slightly more programs
5	Chi-squared	Language × Budget	χ²=250.04, p<10⁻⁵⁴	—	Russian-medium has 5.6× higher budget rate than English
6	Spearman	Country Volume vs Budget Rate	ρ=0.416, p=0.0006	Moderate	High-volume countries tend to have higher budget rates
7	Chi-squared	Region × Budget	χ²=1084.15, p<10⁻²²³	V=0.543	Region is the strongest predictor of budget status
8	Z-test	CIS vs Non-CIS Budget Rates	z=29.55, p≈0	42.3pp gap	49.4% vs 7.1% — the single largest effect

Key Statistical Findings

1. Region is the strongest predictor of budget status (Cramér's V = 0.543 — a large effect). All other variables are partially or fully confounded by regional patterns.

2. CIS membership explains most of the variance in budget acceptance. The 12.7× odds ratio and 42.3 percentage-point gap dwarf all other effects.

3. Language preference is confounded — the Russian-medium budget advantage (29.6% vs 5.3%) disappears when controlling for CIS status, since CIS applicants both prefer Russian AND receive more budget spots.

4. Program count effects are small — the statistically significant Mann-Whitney result (p=0.0005) reflects a trivial mean difference (1.91 vs 1.79 programs) that is again confounded by CIS status.

---

🔑 Major Conclusions

1. The CIS corridor dominates: Russia's bilateral education agreements create a two-tier system where CIS nationals receive budget funding at 7× the rate of other international students.

2. Geographic concentration is extreme: Gini = 0.799 — the top 5 countries contribute nearly half of all applications.

3. Degree-level patterns are consistent: Budget rates decrease with degree level (Bachelor > Master > Specialist), and program specificity increases (PhD applicants are most focused).

4. SPbU's English programs drive diversity: English-taught programs (especially Management) attract the most geographically diverse applicant pool.

5. Iran's unique corridor: Iranian applicants almost exclusively target dental/medical Specialist programs — a highly focused migration pathway.

---

🛠 Data Source & Methodology

Data scraped from three SPbU sources (April 2026):

• cabinet.spbu.ru: Full applicant lists (UID, GUID), competition groups (UID, Country, Programs)
• abiturient.spbu.ru: Olympiad result PDFs with green-highlighted budget winners

The scraper (scrape_spbu.py) uses requests + BeautifulSoup for HTML parsing and PyMuPDF for PDF analysis. Status is determined by detecting green background fill rectangles (RGB ≈ 0.714, 0.843, 0.659) in the PDF files.

---

📜 License

Dataset is derived from publicly available government admissions data published by Saint Petersburg State University.