Poverty, a significant indicator of regional development disparities, exhibits significant heterogeneity within Turkey's spatial and socio-economic structure. Recognizing that this structure is not static and undergoes transformation over time, this research aims to analyze the structural transformation of regional poverty profiles during the 2014-2024 period and the new spatial clustering patterns emerging from this transformation. In this context, the 10-year change and clustering dynamics of regional poverty indicators, based on equivalised household disposable income, across the 26 NUTS Level 2 regions, have been examined.

The study's methodological framework is established to model the structural change in regional poverty profiles in Turkey during the 2014-2024 period and the dynamics of this change. Accordingly, the Gustafson-Kessel (GK) fuzzy clustering algorithm, which offers a flexible approach, has been preferred. This method is capable of effectively modeling the complex and transient profiles of regions, as well as clusters with different geometric properties. The analysis was applied separately to the 2014 and 2024 datasets to comparatively examine the 10-year transformation in the regional poverty map. The analysis utilized regional poverty indicators (poverty threshold, number of poor, and poverty rate) provided by TURKSTAT (Turkish Statistical Institute), based on equivalised household disposable income. To eliminate the effects of different scales among variables and enhance analytical reliability, normalization was achieved by applying z-score standardization to the data.

The empirical findings indicate that a significant structural change occurred in Turkey's regional poverty dynamics between 2014 and 2024. The results reveal that the traditional regional inequality map observed in 2014 was significantly reshaped by 2024.

The analysis shows that in 2014, metropolitan regions (TR10-Istanbul, TR51-Ankara) were part of the cluster with relatively lower poverty rates, while Central, Eastern, and Southeastern Anatolia regions were grouped in clusters with higher poverty rates. Conversely, the 2024 analysis demonstrates a noteworthy change in this structure. Specifically, metropolitan regions are positioned in a cluster with relatively higher poverty rates in 2024, whereas Eastern, Southeastern, and Central Anatolia regions are located in a cluster with relatively more favorable poverty rates.

One of the methodological findings of the study is the success of the Gustafson-Kessel (GK) algorithm in modeling these dynamic transitions. The fuzzy approach of GK, by allowing regions to belong to multiple clusters with 'membership degrees,' has unveiled the changes in these regional profiles with greater nuance compared to conventional methods. The statistical nature of this structural change is also supported by cluster validation metrics. While the 2014 cluster structure (PC: 0.7907, CE: 0.351) indicated a more "fuzzy" and transient profile, the 2024 structure (PC: 0.9531, CE: 0.083) demonstrates a much "clearer" and "sharper" clustering. This situation suggests that within 10 years, the inter-regional poverty divide has not only shifted in direction but also that this new regional structure has become more statistically pronounced.