This paper applies spectral clustering (with Eigen-Gap selection and Silhouette validation) to 2023 import, export and net-export data (raw and exchange-rate adjusted) for Türkiye's 81 provinces, finding an optimal 3-cluster partition (Low/Medium/High) and revealing pronounced regional disparities — e.g., İzmir has the highest net-export performance while İstanbul records the lowest net-export values; exchange-rate adjustments materially change some province rankings .
| Variables | Description | Mean | Mdn. | S.D. | Min. | Max. | Source |
|---|---|---|---|---|---|---|---|
| Export | Thousand US $ | 3,153,558 | 304,126 | 14,251,811 | 139 | 127,097,589 | TurkStat |
| Imports | Thousand US $ | 4,008,624 | 152,298 | 152,298 | 213 | 203,362,138 | TurkStat |
| Net exports | Thousand US $ | -855,065 | 30,170 | 30,170 | -76,264,549 | 4,245,678 | TurkStat |
| Export-adjustment | Thousand US $ | 136,213 | 13,183 | 13,183 | 5 | 5,503,134 | TurkStat |
| Import-adjustment | Thousand US $ | 175,134 | 6,701 | 6,701 | 10 | 8,882,201 | TurkStat |
| Net exports-adjustment | Thousand US $ | -38,920 | 1,323 | 1,323 | -3,379,066 | 183,522 | TurkStat |
引言与报告概览(元数据与总体要点)
逐节深度解读
图表深度解读(每幅图 / 表逐一分析并展示图像)
表 1(Descriptive Statistics)
图 1:Eigen-Gap Method Results
图 2:Silhouette Coefficient Method Results
图 3:Spectral Clustering Results(六幅省级地图:Export / Export-adjustment / Import / Import-adjustment / Net exports / Net export-adjustment)
估值分析
风险因素评估(报告识别的局限与作者说明)
审慎的批判性视角与细微差别
结论性综合(综合关键发现与图表洞见)
建议(对后续研究或报告补充)
总结(两句精要)
2512.04466.pdf