Management and Future Technologies | Volume 2, Issue 2 | № 006

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 bog‘liq bo‘lgan innovatsion jarayonlarni alohida guruhga ajratish imkonini beradi. Xuddi shunday, ishlab chiqarishning texnik darajasining sifat jihatidan o‘sishi bilan bog‘liq bo‘lmagan boshqa barcha sohalarni aniqlash mumkin. Shunday qilib, joriy ishlab chiqarish va ekstensiv kengaytirilgan takror ishlab chiqarish sanoat siyosatining bir yo‘nalishini, intensiv kengaytirilgan takror ishlab chiqarish esa boshqasini tashkil etadi. Ushbu bo‘linish sanoat siyosatida ikkita yo‘nalishni ajratish imkonini beradi: 1) mavjud ishlab chiqarish va uning keng rivojlanishi sharoitida amalga oshiriladigan va 2) ishlab chiqarishni intensivlashtirish, ya'ni innovatsion sanoat siyosatini amalga oshirish bilan bog‘liq. Taklif etilayotgan yondashuv nafaqat innovatsion sanoat siyosatini boshqarish ob'ekti sifatida aniqlash va ajratish, balki boshqaruvning tashkiliy shakllarini shakllantirish bo‘yicha qarorlar qabul qilish imkonini beradi. Ikkinchisi, bir tomondan, ishlab chiqarishning ilmiy-texnikaviy darajasining sifat jihatidan o‘sishi bilan bog‘liq bo‘lgan sanoat siyosatining sohalari bilan, ikkinchi tomondan, mavjud ishlab chiqarish va ishlab chiqarishning ilmiy-texnik rivojlanishining an'anaviy darajasida rivojlanishi mumkin bo‘lgan sohalar bilan shug‘ullanishi mumkin. Mamlakat neft sanoatida soliqqa tortish tizimi bozor munosabatlariga o‘tgandan so‘ng doimiy ravishda sezilarli o‘zgarishlarga duch keldi. Davlat soliq tizimini isloh qilishi va ijara to‘lovlarini yig‘ishi kerak, chunki amaldagi soliq tizimi samarasiz. O‘zbekistonda so‘nggi paytlarda differensial ijara haqini olib qo‘yish masalasi juda dolzarb bo‘lib qoldi. Ijara to‘lovlarini undirish tizimining yo‘qligi neft kompaniyalari tomonidan ishlab chiqarishni jadal rivojlantirish uchun mablag‘ sarflashga intilmayotgan ortiqcha foydaning to‘planishiga yordam beradi, bu esa innovatsion siyosatga salbiy ta’sir ko‘rsatadi, ishlab chiqarish rivojlanishini sekinlashtiradi va chet elga kapital eksportini rag‘batlantiradi. Neft sanoati korxonalarida boshqaruv tizimini tubdan takomillashtirish, birinchi navbatda, innovatsion rivojlanishni strategik rejalashtirish jarayonini rivojlantirish zarur. Ishlab chiqarishning ilmiy-texnik darajasini oshirish bo‘yicha kontseptual ko‘rsatmalar mavjud emas. Bunday hujjatlar nafaqat kompaniya darajasida, balki butun neft va gaz sanoati darajasida ham mavjud emas. Neft-gaz kompleksini innovatsion rivojlantirish strategiyasi mamlakatni ijtimoiy-iqtisodiy rivojlantirish strategiyasining bir qismi sifatida ishlab chiqilishi kerak. Ishlab chiqarishning faoliyat yuritishi va rivojlanishi tizimi innovatsion siyosat asosida qurilishi kerak. (Uzakov[3]) Bizning fikrimizcha, innovatsion siyosat quyidagi asosiy yo‘nalishlarni o‘z ichiga olishi kerak: - Birinchidan, davlat darajasidagi maqsadli sozlashlarga asoslanishi kerak bo‘lgan ijtimoiyiqtisodiy yo‘nalishlarni belgilash; - Ikkinchidan, Neft korxonalarini rivojlanishining prognoz parametrlarini aniqlash; - Uchinchidan, Neft korxonalari faoliyatining barcha sohalarida innovatsion rivojlanishining aniq parametrlarini belgilash; - To‘rtinchidan, Neft korxonalarini rivojlantirishni moliyaviy qo‘llab-quvvatlash manbalarini aniqlash; 241

Volume 2, Issue 2 242 June 2025 Management and Future Technologies journal.umft.uz - Beshinchidan, innovatsion siyosatni amalga oshirish mexanizmlarini ishlab chiqish. Boshqaruv tizimida neft konlarini innovatsion faoliyatini strategik boshqarish organini yaratish zarur, bu esa mamlakat milliy iqtisodiyotining o‘sishi uchun resurs salohiyatidan maksimal darajada samarali foydalanish va mahsulot tannarxini pasaytirish imkonini beradi. Ushbu bo‘lim innovatsion jarayonlarga alohida e'tibor qaratilishi bilan ishlab chiqarishni rivojlantirishni joriy rejalashtirish bo‘limidan sifat jihatidan farq qilishi kerak. Mamlakatda innovatsion siyosatni boshqarishning mustaqil tashkiliy shakllarini ham yaratish zarur. Innovatsion faoliyatni strategik rejalashtirish organining asosiy maqsad va vazifalari quyidagilardan iborat bo‘lishi kerak: - Strategik maqsadlar va rivojlanish rejalarini ishlab chiqish; - Iste'molchilarning ustuvorliklarini aniqlash; - Yangi mahsulotlarni yaratish, joriy etish va bozorga chiqarish; - Bozor sharoitiga qarab talab qilinadigan ishlab chiqarish hajmini aniqlash; - Moliyaviy resurslarning ijobiy hajmini aniqlash va ulardan belgilangan vazifaga erishish uchun foydalanish; - Raqobatbardosh mahsulotlar yaratish va texnologiyani takomillashtirish maqsadida ilmiytexnikaviy siyosatni ishlab chiqish, amalga oshirish va amalga oshirish; - Mehnat resurslarini takror ishlab chiqarishning zarur darajasini ta'minlash; - Ishlab chiqarish samaradorligini oshirish uchun zaxiralarni aniqlash; - Moddiy, mehnat va moliyaviy resurslardan foydalanish samaradorligini doimiy oshirish talablari bilan bir qatorda boshqaruv tizimini takomillashtirish; - Korxona faoliyatining ekologik xavfsizligi; - Kompaniyalarda ilmiy-tadqiqot ishlarini kengaytirish va chuqurlashtirish; - Kompaniyaning integratsiya faoliyatini kuchaytirish; - Yanada zamonaviy texnologiyalarni joriy etish; - Korporativ rivojlanish omillarini aniqlash (ekstensiv, intensiv,narxning oshishi); - Turli darajadagi yangi texnologiyalarni joriy etish hisobiga xarajatlarni minimallashtirish; - Moliyaviy va moddiy resurslardan samarali foydalanish; - Samaradorlik mezonlariga javob bermaydigan aktivlarni chiqarish; - Moliyaviy va ishlab chiqarishni rejalashtirish samaradorligini oshirish; - Moliyaviy resurslarni birlashtirish, optimallashtirish.

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 Maqolada strategik rejalashtirish funksiyalarini ishlab chiqarishni ekstensiv yoki intensiv rivojlantirishga yo‘naltirilganligiga qarab farqlash zarurligi ta'kidlangan. Birinchisi hamma narsani o‘z ichiga oladi, foyda keltiradigan narsa, ikkinchisi - innovatsion rivojlanishga asoslangan ilmiytexnik darajani sifat jihatidan oshirish. Neft sektori uchun soliqqa tortish tizimi hozirgi vaqtda ishlab chiqarishning ilmiy-texnik darajasining o‘sishiga va kompaniyalarning investitsiya resurslarining ko‘payishiga yordam bermaydi, agar neft narxi tushib qolsa, kompaniyalar va umuman sanoat rivojlanishiga salbiy ta'sir ko‘rsatishi mumkin. Agar biz innovatsion jarayonlarni amalga oshirish va ijara daromadidan foyda o‘sishi nisbatini oladigan bo‘lsak, vertikal integratsiyalashgan neft kompaniyalarining yangi texnologiyaga salbiy munosabatining haqiqiy manzarasini ko‘ramiz. Ijara daromadlari sanoatning innovatsion rivojlanishidan olinadigan qo‘shimcha daromadlardan bir necha baravar yuqori. Bundan tashqari, neft qazib olishning innovatsion jarayonlariga qo‘shimcha investitsiyalar uchun uzoq muddatli to‘lovlarni ta'kidlash kerak. Ular xarajatlarni kamaytirish imkoniyatlarining cheklanganligi va kapital xarajatlarning katta hajmlari bilan bog‘liq. Bunday xarajat va foyda nisbati bilan neft kompaniyalari yuqori narxlarga, xom neft uchun eksport siyosatiga va sanoatni rivojlantirishning keng usullariga intilishlari ajablanarli emas. Hozirgi vaqtda neft kompaniyalaridan ortiqcha daromadlarni undirishning samarali tizimi mavjud emas, chunki neft solig‘i neft konlarida turli ishlab chiqarish sharoitlarida differentsial rentani hisobga olish mexanizmini ta'minlamaydi. O‘zbekistonda soliq yukining asosiy ulushi 3 turdagi soliqlarga to‘g‘ri keladi: foydali qazilmalarni qazib olish solig‘i, neftga eksport bojxona to‘lovlari va daromad solig‘i. Ularning neft kompaniyalari tomonidan to‘langan soliqlarning umumiy miqdoridagi umumiy ulushi 2023 yilda taxminan 65-70 foizni tashkil etdi. Tadqiqot davomida foydali qazilmalarni qazib olish solig‘ini joriy etishning salbiy natijalarini aniqlash mumkin edi, ularning asosiylari quyida keltirilgan. 1.Yirik depozitlarda ortiqcha foydaning katta hajmlari soliqqa tortilmasdan qoladi. 2.Soliq solinadigan foydaning salmoqli qismini past rentabelli sohalar tashkil qiladi. Ko‘pincha, foydali qazilmalarni qazib olish solig‘ining ta'siri tufayli ular ishlashni to‘xtatadilar. 3.Ishlash muddatlari uzoq bo‘lgan va neftni olish darajasi past bo‘lgan konlar ham o‘z faoliyatini to‘xtatadi. Ijara daromadi yer osti boyliklari egasi foydasiga qayta taqsimlanishi kerakligi sababli, davlat haqli ravishda neft kompaniyalaridan turli mexanizmlardan foydalangan holda ijara haqini iloji boricha ko‘proq olishga intiladi, ya'ni. soliqlar va boshqa majburiy to‘lovlarni belgilash orqali. Biroq bunday intilishlar qazib oluvchi sanoatni rivojlantirish, ayniqsa, ularning investitsion faolligi hisobiga kelib chiqmasligi kerak. Kon qazib olish tarmoqlarida barqaror rentabellik darajasini ta'minlash uchun konlarni o‘zlashtirishning vaqt omili katta ahamiyatga ega bo‘lgan sharoitlarda davlat konlarni o‘zlashtirishning texnologik xususiyatlarini hisobga olishi va tabiiy renta olishning moslashuvchan mexanizmlaridan foydalanishi kerak. Neft qazib olishda soliqlar davlat siyosatining asosiy vositalari bo‘lib, ular yer qa’ridan foydalanish samaradorligini, sohaning innovatsion va 243

Volume 2, Issue 2 244 June 2025 Management and Future Technologies journal.umft.uz investitsiya salohiyatini oshirishga xizmat qilishi kerak. Neft sanoatidagi ortiqcha foyda renta xarakteriga ega bo‘lib, barcha sohalarda, barcha xilmaxilligida namoyon bo‘lganligi sababli, ularni barcha kompaniyalar uchun yagona soliq stavkasidan foydalangan holda olish qo‘pol xato bo‘lib tuyuladi. Ijara xarakteridagi ortiqcha daromadga soliq stavkasini farqlash kerak. Ushbu xulosa neft qazib olish va neftni qayta ishlash sanoati faoliyat yuritadigan va rivojlanayotgan turli xil sharoitlarni tahlil qilishdan kelib chiqadi (1 va 2-rasmlarga qarang). (Uzakov[6]) 1-rasm. Konlarning renta ortiqcha daromadlariga soliq stavkalarini tabaqalashtirish sabablarining o‘zaro bog‘liqligi. Soliq stavkalarini differentsiallashtirish jarayonida kengaytirilgan takror ishlab chiqarish turiga ekstensiv yoki intensivligiga qarab stavkalar hajmini aniqlash katta ahamiyatga ega. Soliq siyosati ishlab chiqarishning innovatsion rivojlanishini rag‘batlantirishi va ishlab chiqarishning ilmiy-texnik darajasini oshirishi kerak. Neft korxonalarida ishlab chiqarish apparatining, ayniqsa, jismoniy eskirish 80% ga yetadigan neftni qayta ishlash zavodlarining hozirgi holatini hisobga olgan holda, zavodlar, agar ular ishlab chiqarishni rivojlantirish natijasi bo‘lsa, foyda solig‘idan ozod qilinishi kerak. Ushbu qaror muvozanatli yondashuvni talab qiladi. Agar davlat korxonalarni daromad solig‘idan ozod qilsa, u holda bu moliyaviy resurslardan foydalanish sifatini, shu jumladan musodara qilishgacha bo‘lgan nazoratni amalga oshirishi shart. Sanoatning ishlab chiqarish apparati tez orada milliy miqyosdagi qiyinchiliklarga duch kelishi mumkin bo‘lgan paytda neft kompaniyalarining ortiqcha daromadlarini shubhali xarajatlarga ishlatishni to‘g‘ri deb bo‘lmaydi.

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 245 2-rasm. Konlardan qazib olish hajmlari va quduqlarning debetiga qarab imtiyozli soliq stavkalarining shakllanishi. 2-rasmda imtiyozli soliq stavkalarining shakllanishi aks ettirilgan. R - Ordinatalar o‘qida qazib olish hajmi dinamikasi, Q – debet, N soliq stavkasi, G - abssissalar o‘qida, O - konining ishlash davomiyligi, natural ifodadagi mahsulot, sotilgan mahsulot narxi aks ettiriladi. R - egri chizig‘i A nuqtadan C - nuqtagacha konni qazib olishning boshlanishi va oxirigacha bo‘lgan uglevodorodlarni qazib olish hajmi jarayonini tavsiflaydi. B - nuqta maksimal qazib olish hajmiga to‘g‘ri keladi. Diagramma shuni ko‘rsatadiki, qazib olishning boshlang‘ich va yakuniy bosqichlari soliqlar bo‘yicha alohida yondashuvni talab qiladi. Q - egri chizig‘i quduqlar debitining dinamikasini tavsiflaydi, bu ham imtiyozli shartlarni talab qiladi. Biroq, B - nuqtadan C - nuqtagacha debetning o‘sishi innovatsion omillar ta’sirining natijasidir. Masalan, gorizontal va gorizontal-tarmoqlangan quduqlar yordamida konlarni ishlatish texnologiyasi joriy debitlarni bir necha barobar oshiradi. Innovatsion omillardan foydalanish, muallifning fikriga ko‘ra, Soliq kodeksidagi soliq yukini engillashtirishi yoki olib tashlashi kerak. Bizningning fikrimizcha, ijara to‘lovidan ortiqcha daromadlarni davlat va soliq to‘lovchi o‘rtasida taqsimlash muhim masaladir. Ushbu muammoning universal echimi NC ga intensiv kengaytirilgan ko‘paytirishni amalga oshirish imkonini beradi. G‘arb rivojlangan mamlakatlar tajribasi shuni ko‘rsatadiki, ortiqcha foydaning 20-30% MKlarga samarali innovatsion siyosatni amalga oshirish imkonini beradi. Bu tub chora-tadbirlarni xo‘jalik amaliyotiga joriy etish juda ehtiyotkorlik bilan amalga oshirilishi kerak. Har doim esda tutish kerakki, NTCning respublika byudjetdagi ulushi 10-12%, eksportda - 45%, valyuta tushumlari miqdorida - 45%. Shu sababli, NGK boshqaruv tizimining sifat parametrlarini o‘zgartirishda katta e'tibor talab etiladi. Samarali innovatsion siyosatni amalga oshirish uchun soliq imtiyozlari tizimini joriy etish taklif etilayotgan chora-tadbirlarni amalga oshirish uchun zaxira bo‘lishi mumkin. Imtiyozli soliqqa tortish soliq solinadigan bazani kamaytirish yoki soliq stavkalarini pasaytirish yoki soliq to‘lovlari miqdorini kamaytirish orqali amalga oshirilishi mumkin. Asosiysi, ishlab chiqarishni texnik yangilashning har qanday turini emas, balki mahsulot birligi tannarxini pasaytiradigan va kapital xarajatlar o‘rtacha foyda darajasiga mos keladigan narsalarni rag‘batlantirishdir. Shuni yodda tutish kerakki, taklif etilayotgan radikal chora-tadbirlar soliq organlaridan ortiqcha foydani musodara qilmasdan turib kutilgan natijani bera olmaydi. Shuning uchun bu choralar bir vaqtning o‘zida va kechiktirmasdan

Volume 2, Issue 2 246 June 2025 Management and Future Technologies journal.umft.uz amalga oshirilishi kerak. Xulosa Mamlakat neft sanoatining G‘arb texnologiyalariga qaramligini kamaytirish uchun davlat va biznes tomonidan mahalliy mashinasozlikni rivojlantirish bo‘yicha strategik dastur ishlab chiqish zarur. Har bir neft korxonalari ishlab chiqarish faoliyatining asosiy turlarini (mijozlarning funksiyalari) o‘zining ilmiy-texnik ta'minotiga ega bo‘lishi kerak. Texnik jihatdan qayta jihozlash va rekonstruksiya qilish sanoatni rivojlantirishning eng muhim yo‘nalishlari bo‘lishi kerak. Bu ishlab chiqarish tuzilmasini sezilarli darajada yaxshilaydi, mahsulot hajmining jadal o‘sishini ta'minlaydi. Bizning fikrimizcha, Neft korxonalari boshqaruv tizimida innovatsion rivojlanish muammolarini aniqlash va ularni ixtisoslashtirilgan tashkilotga buyurtma berish orqali hal qilishga qaratilgan innovatsion rivojlanish bo‘limi ajratish kerak - yanada ilg‘or texnologiyalarni joriy etish orqali qazib olish, qayta ishlash, tashish xarajatlarini kamaytirish. Innovatsion rivojlanish bo‘limining vazifalari quyidagilardan iborat bo‘lishi kerak: ishlab chiqarishning texnik darajasini oshirish; rentabellik, aktivlar rentabelligi va mehnat unumdorligi darajasini oshirish; kapital qo‘yilmalar samaradorligini oshirish; xarajatlarni kamaytirish; asbob-uskunalar sifatini va uning ishlash ishonchliligini oshirish; atrof-muhit muhofazasini yaxshilash va uning ifloslanishining oldini olish; ilg‘or texnologiyalarni doimiy ravishda izlash maqsadida ilmiy tadqiqotlar hajmini oshirish; innovatsion rivojlanishni boshqarishning texnik vositalari va tashkiliy shakllarini takomillashtirish.

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 Adabiyotlar 1. 1.Neft va gaz korxonalarini innovatsion rivojlantirish” xorijiy mamlakatlar tajribasi. Ma’ruza tezisi // “Korxonalarni strategik rejalashtirish va rivojlantirish” Oltinchi Butunrossiya simpoziumi, 2005 yil 12-13 aprel. 3-bo‘lim “Rossiya va xorijiy korxonalarda strategik rejalashtirish tajribasi”. - M.: 01.05 . 2. 2. Yakubov M.S, Uzakov B.M, Xoshimov B.M, Neft qayta ishlash jarayonini matematik modellashtirish va avtomatlashtirish haqida. // Polsha.2024. No. jild. 48 (2024): Miasto Przyszłości 213-216-betlar. 3. 3.Uzakov B.M, Xoshimov B.M, O‘zbekiston neft-gaz tomonidan ishlab chiqarishni moliyalashtirish boy‘icha xorijni o‘rganish. //Muhammad al-Xorazmiy nomidagi TATU Farg‘ona filiali “Al-Farg‘oniy avlodlari” elektron ilmiy jurnali ISSN 2181-4252 Tom: 1 | O‘g‘il: 1 |10-19 betlar| 2024 yil. 4. 4.Uzoqov B.M. Xoshimov B.M. Distillash ustunining sifat ko‘rsatkichlarining virtual analizatorlari modellarini aniqlash usullarini o‘rganish. // Muhammad al-Xorazmiy nomidagi TATU Farg‘ona filialining “Al-Farg‘oniy avlodlari” elektron ilmiy jurnali ISSN 2181 -4252 Jild: 1 | Nashr: 1 |2024 yil. 5. 5.Uzoqov B.M. Texnologik jarayonlarni operativ boshqarish modellarini texnik-iqtisodiy ko‘rsatkichlar bo‘yicha moslashtirish // Muhammad al-Xorazmiy nomidagi TATU Farg‘ona filialining “Al-Farg‘oniy avlodlari” elektron ilmiy jurnali ISSN 2181 -4252 Jil: 1 | Muammo: 2 | 2023 yil 6. 6.Yoqubov M.S., Uzoqov B.M., Mahsulot sifati ko‘rsatkichlari bo‘yicha tezkor boshqaruv ostida neftni qayta ishlash jarayonlarini modellashtirish. // Muhammad al-Xorazmiy nomidagi TATU Farg‘ona filialining “Al-Farg‘oniy avlodlari” elektron ilmiy jurnali. ISSN 2181 -4252 jild: 1 |Iss: 1 | 2023 yil. 7. 7.Uzoqov B.M Neftni qayta ishlash korxonalarining boshqaruv tizimini takomillashtirish faoliyati // Yevropa iqtisodiyot, moliya va biznesni rivojlantirish jurnali ISSN (E): 2938-3633 2-jild, 5-son, may - 2024 yil 8. 8.M. Yoqubov , B.M. Uzoqov Modellashtirish jarayonlar neftni qayta ishlash da operativ boshqaruv tomonidan ko‘rsatkichlar sifat mahsulotlar // Ilmiy- texnik jurnal (STJ FerPI, FarPI ITZH , NTZH FerPI , 2023 yil, T.27 . maxsus soni 2) (40-46 betlar) 9. 9.Uzakov B.M Neft maxsulotlarini ekstraksiyalash jarayonlarini modellashtirish va boshqarish usullarini tahlili./ materiallar. Farg‘ona 4-5-may 2023 yil. (459-462 betlar) 10. 10.Yakubov M.S, Uzakov B.M, Xoshimov B.M, Farg‘ona neftni qayta ishlash zavodi uchun avtomatlashtirilgan matematik modeli va algoritmlash jadvalini hisoblash vazifalari// Muhammad alXorazmiy nomidagi TATU Farg‘ona filiali “Al-Farg‘ona filiali. avlodlari” elektron ilmiy jurnali ISSN 2181-4252 Jild: 1 | O‘g‘li: 2 | 2024 yil 11. 11.Uzakov B.M. Neftni qayta ishlash jarayonini boshqarish IS tuzatish // Muhammad al-Xorazmiy nomidagi TATU Farg‘ona filiali “Al-Farg‘oniy avlodlari” elektron ilmiy jurnali SN 2181-4252 Tom: 1 | O‘g‘li: 2 | 2024 yil 12. 12.Uzakov BM Neft ishlab chiqarishni ekstraksiyalash jarayoni modellashtirish va boshqarish usullarini tahlili // Muhammad Al-xorazmiy nomidagi Toshkent axborot texnologiyalari Universiteti Farg‘ona filiali Aniq va tabiiy fanlarni rivojlantirishda rejalashtirishtexnologiya Ularning o‘rni: muammolar va innovatsion yechimlar mavzusidagi xalqaro ilmiy-texnik konferensiya materiallari to‘plami 3-qism (556-558 betlar) 247

Volume 2, Issue 2 June 2025 248 Authors Khidirova Charos Department of Computer Systems, Tashkent University of Information Technologies named after Muhammad al-Khwarizmi, Uzbekistan; khidirova@tuit.uz Abdullakhujaev Saidislombek Department of Computer Systems, Tashkent University of Information Technologies named after Muhammad al-Khwarizmi, Uzbekistan; saidislomabdullakhodjaev@gmail.com Correspondence: khidirova@tuit.uz Copyright: © 2025 by the authors. Ushbu maqola Creative Commons Attribution (CC BY) litsenziyasi shartlari asosida tarqatiladigan ochiq foydalanish maqolasi hisoblanadi (https://creativecommons.org/ licenses/by/4.0/). Management and Future Technologies journal.umft.uz COMPARATIVE ANALYSIS OF DATA MINING METHODS FOR EARLY DETECTION OF ONCOLOGICAL DISEASES A bstract: This paper presents a comprehensive review and comparative analysis of modern machine learning algorithms and intuitionistic fuzzy set methods applied to the early diagnosis and risk predection of oncological diseases, with a particular focus on cervical cancer. The study evaluates classical and ensemble models, including Random Forest, XGBoost, AdaBoost, SVM, Decision Trees, and deep learning architectures, analyzing their accuracy, interpretability, and clinical applicability. Special emphasis is placed on the application of IFS theory, utilizing Euclidean and Hamming distances, as a robust tool for medical diagnosis. Experimental findings from over 60 sources demonstrate that ensemble methods (Random Forest, AdaBoost) and hybrid approaches (CatBoost, Majority Voting) achieve the highest predictive accuracy, with Random Forest reaching 100% in cervical cancer classification tasks. The stability and reproducibility of the IFS-based diagnosis method are also highlighted, confirming its potential for integration into intelligent clinical decision support systems. This study lays the foundation for developing personalized medical solutions based on machine learning and fuzzy logic techniques. Keywords: machine learning, data mining, disease prediction, early cancer detection, Random Forest, ensemble methods, intuitionistic fuzzy sets, Euclidean distance, Hamming distance, cervical cancer, intelligent medical systems.

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 Introduction The early detection of oncological diseases remains one of the most critical challenges in modern healthcare. Timely identification of cancer significantly increases survival rates, reduces treatment costs, and enhances the quality of life for patients. In this context, the integration of artificial intelligence, particularly machine learning and data mining methods, has opened new avenues for improving diagnostic accuracy and clinical decision-making. With the exponential growth of medical data – from electronic health records and imaging to genomic and laboratory results – there is a growing need for intelligent systems capable of extracting meaningful patterns and predictive features. Data mining algorithms, encompassing classical statistical models, supervised and unsupervised machine learning techniques [17-19], as well as ensemble and deep learning architectures, have demonstrated remarkable performance in diagnosing complex diseases such as cervical, breast, lung, and skin cancers. Recent studies have highlighted the potential of ensemble methods like Random Forest, XGBoost, and AdaBoost to achieve high diagnostic precision. At the same time, interpretable models such as Decision Trees and Support Vector Machines offer transparency essential for clinical adoption. Furthermore, the use of fuzzy logic, particularly Intuitionistic Fuzzy Sets (IFS), has proven effective in handling uncertainty and imprecision inherent in medical data, thereby enhancing the robustness and generalizability of diagnostic models. This paper aims to present a comprehensive comparative analysis of modern data mining techniques for the early detection of oncological diseases. It synthesizes findings from over sixty scientific sources, evaluates model performance across multiple clinical tasks, and highlights the role of IFS-based approaches in medical diagnostics. The insights obtained contribute to the design of intelligent decision support systems that support personalized and evidence-based medicine. Comparative Study of Data Mining Approaches for Oncological Diagnostics The rapid development of machine learning and data mining technologies has significantly expanded their applications in the field of medical diagnostics, particularly in oncology. In this study, a detailed comparative analysis of various machine learning algorithms was conducted with the aim of identifying the most effective methods for the early detection of oncological diseases. The evaluation included classical algorithms such as Support Vector Machine (SVM), Decision Tree (DT), and Logistic Regression (LR), as well as advanced ensemble methods like Random Forest (RF), XGBoost, AdaBoost, and CatBoost. Deep learning architectures, including Convolutional Neural Networks (CNN) and Multilayer Perceptrons (MLP), were also considered. Furthermore, the study examined hybrid approaches and the application of Intuitionistic Fuzzy Sets (IFS) for diagnostic purposes. 249

Volume 2, Issue 2 250 June 2025 Management and Future Technologies journal.umft.uz The performance of these algorithms was assessed based on various clinical datasets covering diseases such as cervical cancer, lung cancer, melanoma, cardiovascular diseases, and diabetes-related oncological risks. Special attention was given to metrics such as accuracy, Area Under the Receiver Operating Characteristic Curve (AUC-ROC), interpretability, robustness to missing data, and clinical applicability. In this section, recent scientific works are reviewed that illustrate the application of these machine learning and data mining methods in practical healthcare tasks. Key findings, strengths, and limitations of each method are discussed, with a particular focus on their potential for integration into intelligent decision support systems for personalized medicine. In the study Nafiseh Ghaffar Nia and his colleagues [1], discuss the application of artificial intelligence methods for disease diagnosis and prediction using machine learning and deep learning algorithms. The following methods are analyzed: Support Vector Machine (SVM), Random Forest (RF), K-Nearest Neighbors (KNN), Decision Tree (DT), Deep Belief Network (DBN), Long ShortTerm Memory (LSTM), and Convolutional Neural Network (CNN). The CNN algorithm achieved an accuracy of 98.91% for COVID-19 and 99.51% for breast cancer. The SVM method demonstrated up to 99.51% accuracy, Random Forest achieved up to 95.58%, K-Nearest Neighbors reached 96.33%, and Decision Tree achieved 93.62%. CNN and SVM algorithms showed the best performance in the tasks of diagnosing oncological and viral diseases. A schematic representation of the main directions of machine learning (ML) is provided: models are divided into three large categories — supervised learning, unsupervised learning, and reinforcement learning — each including a number of widely used algorithms (e.g., Support Vector Machine (SVM), K-means clustering, and others). In the research Kamrul Hasan and his colleagues [2], address the problem of early detection of cardiovascular diseases using machine learning methods with a focus on hyperparameter optimization. The Cleveland dataset (303 records) and the extended Comprehensive dataset (1025 records) were utilized.The following algorithms were applied:Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machine (SVM),Decision Tree (DT), Random Forest (RF), and XGBoost.Hyperparameter tuning was performed using GridSearchCV.The Support Vector Machine (SVM) achieved an accuracy of up to 88%, the Decision Tree (DT) reached 78%, and both XGBoost and Random Forest achieved up to 99% accuracy on the larger dataset.The most effective methods were ensemble techniques — XGBoost and Random Forest. In the study Tewodros Mulugeta and colleagues [3], analyze the impact of hyperparameter tuning on the accuracy of cardiovascular disease prediction using the GridSearchCV method.The following algorithms were tested: Logistic Regression, K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Decision Tree (DT), Random Forest (RF), and XGBoost, using the Cleveland dataset (303 records) and a combined dataset (1025 records).The Support Vector Machine (SVM) achieved an accuracy of 87.91% on the smaller dataset, while XGBoost reached 99.03% on the larger

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 dataset.The Decision Tree (DT) and Random Forest (RF) showed up to 98.05% accuracy.Logistic Regression and K-Nearest Neighbors (KNN) achieved approximately 82%.XGBoost demonstrated the best performance when working with extended datasets. In the study Fangxuan Li and others [4], describe a predictive model for gastroparesis following complete mesocolic excision for colorectal cancer based on clinical and surgical data from 1146 patients.Four algorithms were tested: XGBoost, Random Forest, Support Vector Machine (SVM), and K-Nearest Neighbors (KNN).The evaluation was conducted using k-fold cross-validation, ROC and DCA curves, while interpretability was achieved through SHAP analysis.The XGBoost model demonstrated high accuracy and stability during both internal and external validation. Among all algorithms, XGBoost showed the best results in predicting gastroparesis (Figure 1). Figure 1. ROC, calibration, and DCA curves for evaluating gastroparesis prediction models.[4] Figure 2 illustrates the internal and external validation of the XGBoost model, including a comparison of ROC curves for the training, validation, and test datasets, which enabled an assessment of the model's stability and generalization ability when applied to different samples. 251

Volume 2, Issue 2 June 2025 Management and Future Technologies journal.umft.uz 252 Figure 2. ROC curves of the XGBoost model for training, validation, test sets, and external validation.[4] In the article Jiwon Baek and colleagues [5], present a deep learning model for predicting fellow eye neovascularization in patients with unilateral type 3 macular neovascularization.The study used data from 217 patients, including fundus photographs and OCT images of the fellow eye, processed using the VGG architecture, data augmentation, and transfer learning techniques.The model achieved an AUC of 67.5% and an accuracy of 66.7%, comparable to the performance of experienced ophthalmologists.The deep neural network model delivered the best results, although its accuracy was limited by the complexity of the task and the relatively small dataset. In the scientific work Ying Wang and colleagues [6], present a diagnostic model for active tuberculosis based on plasma immune profiling using 92 biomarkers. The following algorithms were applied: Support Vector Machine (SVM), Decision Tree (DT), Naïve Bayes, Logistic Regression, Linear Discriminant Analysis (LDA), K-Nearest Neighbors (KNN), and Neural Network. The data were collected from 222 patients with various clinical conditions. The Support Vector Machine (SVM) model achieved an AUROC between 0.89 and 0.99 for diagnosing active tuberculosis, exceeding 0.95 in severe cases, and greater than 0.8 in outcome prediction. Among all tested methods, SVM demonstrated the highest accuracy and stability in both prognosis and diagnosis.

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 In the study Jun Chen and colleagues [7], investigate the differentiation of lymphoma and benign lymph node lesions in cases of fever of unknown origin (FUO) using radiomics based on 18FFDG PET/CT imaging.The models were built using Logistic Regression, Support Vector Machine (SVM), Random Forest (RF), and K-Nearest Neighbors (KNN), with feature selection performed via the LASSO method.Training was conducted on data from 143 patients, with testing on a separate cohort of 61 patients.The Random Forest (RF) algorithm demonstrated the best performance, achieving an accuracy of 97.9% and an AUC of 0.998 during training, and an accuracy of 83.9% with an AUC of 0.915 during testing.Other models also showed good performance, but Random Forest proved to be the most accurate tool for lymphoma diagnosis based on radiomic data. In figure 3 (a) and (b) present the ROC curves of four models on the training and test datasets, respectively (Figure 3). Figure 3. ROC curves of four models.[7] In the research work Benjamin M. Schmidt and others [8], present machine learning models for the early prediction of cervical cancer based on electronic health records (EHR) from a Swedish clinical database. The study analyzed data from 1321 patients and 16,212 control records, encompassing diagnoses, prescriptions, procedures, laboratory results, and clinical text entities extracted using Long Short-Term Memory (LSTM) networks and Named Entity Recognition (NER). A hierarchical feature structure was employed to enhance model performance. Among the tested models, Random Forest achieved the best outcomes: an AUC of 0.97 one day before diagnosis and 0.70 one year before diagnosis, with an accuracy of up to 0.92 and a specificity exceeding 97%. The Random Forest model demonstrated excellent accuracy and interpretability, particularly when working with textual data and hierarchical feature representations (Figure 4). 253

Volume 2, Issue 2 June 2025 Management and Future Technologies journal.umft.uz 254 Figure 4. Comparison of AUC and misclassification rates between the old integer score system, the new multivariate logistic regression model, and the new integer score system.[8] In the paper H. Chen and other researchers from Harvard Medical School [9], present the CHIEF artificial intelligence model for the diagnosis, prognosis, and identification of molecular characteristics of oncological diseases based on histopathological images. The model was trained on more than 15 million images and tested on 32 independent datasets from 24 clinics. CHIEF achieved an accuracy of 96% in analyzing biopsy samples and 94% in surgical samples, showing an improvement in accuracy by 8–10% compared to existing solutions and demonstrating high efficiency in detecting BRAF, NTRK1, and EZH2 mutations. The CHIEF model exhibited excellent accuracy and significant potential for application in personalized oncology. In the research work Giuseppe Mercante and colleagues [10], explore the application of machine learning methods to improve the performance of ultrasound (US) imaging in head and neck oncology. The review includes an analysis of modern models such as Support Vector Machine (SVM), Random Forest (RF), Decision Tree (DT), Convolutional Neural Network (CNN), and Neural Networks, used for automatic classification of benign and malignant lesions, tumor segmentation,

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 and sonographic data interpretation. The studies reviewed demonstrate that the use of machine learning models enables classification accuracies of up to 94–96% in diagnosing thyroid, salivary gland, and lymph node cancers. Ensemble methods and deep learning approaches, particularly CNNs, are recognized as the most promising, especially when working with large datasets and pre-annotated samples. Thus, machine learning significantly enhances the diagnostic value of ultrasound imaging in head and neck oncology, especially through the application of deep neural networks (Figure 5). In the article Meenu Gupta and colleagues [11], present a cervical cancer prediction model based on the open UCI dataset, comprising 858 records and 32 risk features. The tested algorithms included Random Forest (RF), Decision Tree (DT), Adaptive Boosting (AdaBoost), Support Vector Machine (SVM), Logistic Regression (LR), and K-Nearest Neighbors (KNN). Ensemble methods — Random Forest, Decision Tree, and Adaptive Boosting — achieved 100% accuracy. The Support Vector Machine reached 99%, Logistic Regression 98%, and K-Nearest Neighbors 95%, although the latter demonstrated low recall (31%). Ensemble algorithms exhibited the best results and highest reliability in early diagnosis tasks. Figure 5. Machine learning methods in ultrasound radiomics for head and neck oncology.[9] In the study by Xue Li et al. [12], lung cancer risk prediction varied by histological subtype. For adenocarcinoma (ADC), the XGBoost algorithm with SNP data achieved an AUC of 0.699 (95% CI: 0.666–0.733), significantly outperforming the baseline logistic regression model (AUC = 0.639, p = 0.009). This supports the idea that genetic markers like CHRNB3 rs4236926 improve ADC prediction when used with machine learning. 255

Volume 2, Issue 2 256 June 2025 Management and Future Technologies journal.umft.uz For squamous cell carcinoma (SCC), the highest AUC (0.833) came from a logistic regression model using epidemiological data and SNPs (p = 0.491 vs. baseline AUC = 0.818). This suggests that in SCC, where factors like smoking and age dominate, traditional models remain more effective than complex ML algorithms. Table 1. Lung cancer risk prediction (ADC and SCC).[12] Model Epidemiologyonly model Method LR XGBoost Full model LR XGBoost Epidemiologyonly model LR XGBoost Full model LR XGBoost AUC Sensitivity 95% CI ADC - Adenocarcinoma 0.606– 0.639 0.632 0.669 0.652– 0.683 0.638 0.714 0.633– 0.666 0.551 0.698 0.666– 0.699 0.739 0.733 SCC - Squamous Cell Carcinoma 0.786– 0.786 0.818 0.847 0.785– 0.814 0.754 0.842 0.805– 0.833 0.745 0.859 0.782– 0.816 0.762 0.847 AUC Specificity p 0.596 Ref 0.666 < 0.001 0.695 0.255 0.567 0.009 0.737 Ref 0.788 0.848 0.805 0.491 0.745 0.893 In the study G. Saravanan, and colleagues [13], propose a hybrid approach for diagnosing skin melanoma using dermatoscopic images and machine learning algorithms. The tested methods included K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Convolutional Neural Network (CNN), Naive Bayes, and the ensemble method Majority Voting. The highest accuracy (99.23%) was achieved by the Majority Voting ensemble model. Among individual algorithms, Naive Bayes (98.21%) and Support Vector Machine (97.93%) demonstrated the best results, while the Convolutional Neural Network (97.88%) showed stable efficiency even with a limited dataset. The ensemble Majority Voting method exhibited excellent performance and high potential for clinical application. In the study Qian Qiu and colleagues [14], present a cancer risk prediction model for patients with type 2 diabetes based on 107 clinical and laboratory indicators selected from 1381 features. Sixteen algorithms were tested, including Logistic Regression (LR), Support Vector Machine (SVM), Naïve Bayes, K-Nearest Neighbors (KNN), Adaptive Boosting (AdaBoost), Random Forest (RF),

Management and Future Technologies journal.umft.uz June 2025 Volume 2, Issue 2 XGBoost, LightGBM, and CatBoost. The best performance was demonstrated by CatBoost, achieving an AUC of 85.2% and an accuracy of 79.6%. LightGBM and Random Forest also performed strongly, with AUC values of 83.6% and 83.3%, respectively. CatBoost delivered the highest balanced efficiency, making it the preferred method for cancer risk prediction based on routine medical tests (Figure 6). Figure 6. ROC curves of the top five classifiers.[14] In the scientific work Tingting Wu and colleagues [15], develop a predictive model for postoperative survival in patients with gastric cancer based on SEER data (11,414 records) and external validation on a Chinese cohort (2846 patients). The study compares deep learning (DL), multitask logistic regression (MTLR), Random Survival Forest (RSF), and the clinical TNM staging system. The deep learning model achieved the highest AUC values: up to 0.82 on the training dataset and 0.77 on the external validation dataset. It also provided superior clinical utility and calibration. The deep learning approach demonstrated excellent predictive performance and generalizability, making it the most promising tool for survival prediction. In the study Chan S. W. and Abdullah L. [16], developed a cancer diagnosis method based on Intuitionistic Fuzzy Sets (IFS) using Euclidean and Hamming distances. The methodology involved forming intuitionistic fuzzy relations between patient symptoms and various types of cancer, including lung, breast, colorectal, nasopharyngeal, and cervical cancers. The resulting data, presented in Table 2, reflected the computed distances between patients and cancer types. A comparison of results using Euclidean and Hamming distances demonstrated consistency, confirming the effectiveness of applying IFS in medical diagnostics. The authors highlighted the stability and reproducibility of the proposed method for determining the probable type of oncological disease (Table 2). 257

Volume 2, Issue 2 June 2025 Management and Future Technologies journal.umft.uz 258 Table 2. Results of IFS Euclidean distance calculations between patients and various types of cancer.[16] Patient’s name Lung Cancer Breast Cancer Colorectal Cancer Nasopharyngeal Cancer Cervical Cancer Fati 0.3844 0.3667 0.1826 0.2769 0.1414 Lam 0.4295 0.4123 0.0816 0.3958 0.1563 Kuz 0.2687 0.2867 0.4269 0.1453 0.3873 Shei 0.1915 0.1453 0.4177 0.2789 0.4082 Hal 0.0943 0.1633 0.4256 0.3018 0.4041 Research results The conducted analysis of scientific articles on the application of machine learning algorithms in medical diagnosis and prognosis demonstrated the high effectiveness of modern models, particularly in the early detection of oncological diseases. The Random Forest (RF) algorithm showed the highest performance, achieving up to 100% accuracy in diagnosing cervical cancer and 99.5% in breast cancer detection. It also demonstrated reliability in survival prediction and lymphoma diagnosis.The Support Vector Machine (SVM) algorithm provided up to 99.5% accuracy in oncology tasks and over 95% accuracy in diagnosing COVID-19 and diabetes, showing strong performance even with limited training data. Despite its simple structure, the Decision Tree (DT) algorithm achieved up to 100% accuracy in individual studies and is noted for its high interpretability, which is particularly valuable for clinical decision support systems. In addition, ensemble and hybrid methods such as XGBoost, AdaBoost, CatBoost, and Majority Voting demonstrated excellent results, especially in melanoma diagnosis and cancer risk prediction tasks. Thus, RF, SVM, DT algorithms, as well as ensemble approaches, demonstrated the best accuracy and applicability for medical AI systems.A particularly important finding is the use of Intuitionistic Fuzzy Sets (IFS) as proposed by Chan S.W. and Abdullah L. [16]. Their method, based on Euclidean and Hamming distances, proved to be stable and reproducible, enabling the effective classification of various cancer types. This approach opens up new opportunities for the integration of fuzzy logic methods into clinical diagnostic systems.However, to enhance the reliability and reproducibility of predictive models, it is necessary to standardize data, perform multicenter validation, and ensure the integration of these solutions into existing clinical protocols (Table 3). Table 3. Performance Comparison of Machine Learning and Intuitionistic Fuzzy Methods in Disease Diagnosis and Prediction

Management and Future Technologies journal.umft.uz June 2025 № AI methods Application Area 1 2 Random Forest (RF) Random Forest Classifier (RFC) Adaptive Boosting (AB) Bernoulli Naive Bayes (BNB) CatBoost (CB) Diabetes prediction, heart disease prediction Heart disease prediction (Cleveland dataset) 3 4 5 6 7 8 9 10 11 12 13 14 Cardiovascular disease prediction Cervical cancer classification Analysis of the presence/absence of medical events Cancer risk prediction Cardiovascular disease classification Heart disease prediction (Cleveland dataset) Cardiovascular disease prediction Risk prediction of postoperative gastroparesis after complete mesocolic excision (validation cohort) Risk prediction of postoperative gastroparesis after complete mesocolic excision (external validation Support Vector cohort) Machine Diagnosis and prediction of active tuberculosis (SVM) Diagnosis of lymphoma in patients with fever of unknown origin (FUO) [4] Diagnosis of lymphoma in patients with fever of unknown origin (FUO) [7] 15 High-dimensional feature classification 16 17 18 19 Disease classification by systems Diagnosis of nodule malignancy Cervical cancer classification Melanoma classification 20 Breast and lung cancer diagnosis 21 22 Decision Tree (DT) Heart disease prediction (Cleveland dataset) Cardiovascular disease prediction Volume 2, Issue 2 Accuracy (AUCROC) 91% (diabetes) 82–98% 84.62% (DatasetI), 98.05% (DatasetII) 100% 61% (per year), 92% (per day) 85% 95% 88–90% 84.62% (DatasetI), 98.05% (DatasetII) 85.70% 72.10% 89–99% 86.60% 93% 12–32% (per year), 80% (per day) 50% 99% 99% 97.93% 83% (oncology), 93.62% (cancer) 97% 61.32% (DatasetI), 97.08% (DatasetII) 259

260 Volume 2, Issue 2 June 2025 № Application Area AI methods 23 24 25 26 27 28 29 Management and Future Technologies Diagnosis and prediction of active tuberculosis Disease classification by systems Cervical cancer classification Deep Belief Medical data analysis for pathology detection Network (DBN) Complement Naive Bayes Handling class imbalance in data (CNB) Multilayer Disease prediction based on symptoms Perceptron (MP) Diabetes and skin disease diagnosis (diabetes) journal.umft.uz Accuracy (AUCROC) 82–93% 50% 100% 88% (pathology diagnosis) 45–85% (depending on the term) 99.17% 30 Cardiovascular disease prediction 31 32 Heart disease prediction Cardiovascular disease prediction 87% 87.91% (DatasetI), 83.44% (DatasetII) 83.90% 70.80% Diagnosis and prediction of active tuberculosis 77–85% 33 34 K-Nearest Neighbors (KNN) 35 36 37 38 39 40 41 42 43 Multivariate Logistic Regression (MLR) Complement Naive Bayes (CNB) Naive Bayes (NB) Lymphoma diagnosis (fever of unknown origin) (FUO) [4] Lymphoma diagnosis (fever of unknown origin) (FUO) [7] Disease classification by systems Cervical cancer classification Melanoma classification 50% 95% 96.11% Preoperative diagnosis / combined models 83–93% 89.10% 93.80% Therapy response prediction Benign/malignant tumor classification 45–85% (depending on the term) 87% 98.21% Heart disease prediction (Cleveland dataset) 82–87% Handling class imbalance in data