MIGRATSION OQIMLARNI PROGNOZLASHDA "KATTA MA'LUMOTLAR" (BIG DATA) VA MASHINANI O'QITISH ALGORITMLARINING O'RNI: JAMOAT XAVFSIZLIGINI TA'MINLASH VA INQIROZLARNI BARVAQT ANIQLASH

Аннотация

Annotatsiya:Ushbu maqolada zamonaviy migratsion jarayonlarni prognozlashda "katta ma'lumotlar" (Big Data) texnologiyalari va mashinani o'qitish (Machine Learning, ML) algoritmlaridan foydalanishning nazariy va amaliy jihatlari tahlil qilinadi. Global miqyosda migratsiya hodisalarining tobora murakkablashib borayotgani, ularning tezkor va noan'anaviy xarakter kasb etayotgani an'anaviy statistik uslublarning (aholi ro'yxati, so'rovnomalar) real vaqt rejimida samarali qaror qabul qilish uchun yetarli emasligini ko'rsatmoqda. Maqolada ijtimoiy tarmoqlar, mobil aloqa operatorlari ma'lumotlari, sun'iy yo'ldosh tasvirlari, qidiruv so'rovlari va moliyaviy tranzaksiyalar kabi noan'anaviy ma'lumot manbalarining migratsion oqimlarni bashorat qilishdagi salohiyati ko'rib chiqiladi. Shuningdek, regressiya modellari, tasodifiy o'rmon (Random Forest), gradient bustlash, chuqur neyron tarmoqlar (LSTM, CNN) kabi mashinani o'qitish algoritmlarining amaliy qo'llanilishi, xalqaro tashkilotlar (IOM, UNHCR, Frontex) tajribasi misolida tahlil etiladi. Maqolaning alohida qismi ushbu texnologiyalarning jamoat xavfsizligini ta'minlash, insonparvarlik inqirozlarini barvaqt aniqlash va O'zbekiston sharoitida qo'llash istiqbollariga bag'ishlangan. Xulosa qismida ma'lumotlar maxfiyligi, algoritmik nosozlik (bias) va institutsional tayyorgarlik borasidagi muammolar hamda ularni bartaraf etish yo'llari bo'yicha tavsiyalar berilgan.

Abstract:This article examines the theoretical and applied dimensions of using Big Data technologies and machine learning (ML) algorithms to forecast contemporary migration flows. The growing complexity and rapid onset nature of global migration events demonstrate that traditional statistical methods — censuses and surveys — are insufficient for real-time decision-making. The article reviews the predictive potential of non-traditional data sources such as social media, mobile network operator records, satellite imagery, search engine queries, and financial transaction data. It further analyzes the practical application of machine learning algorithms — regression models, Random Forest, gradient boosting, and deep neural networks (LSTM, CNN) — through the experience of international organizations such as IOM, UNHCR, and Frontex. A dedicated section addresses the role of these technologies in ensuring public safety, enabling early detection of humanitarian crises, and prospects for application in Uzbekistan. The conclusion offers recommendations regarding data privacy, algorithmic bias, and institutional readiness.