الذكاء الاصطناعي في تحسين المالية وسلسلة التوريد: التحليلات التنبؤية لنمو الأعمال واستقرار السوق في الولايات المتحدة الأمريكية Artificial Intelligence in Financial and Supply Chain Optimization: Predictive Analytics for Business Growth and Market Stability in The USA

المجلة: International Journal of Applied Sciences and Radiation Research، المجلد: 2، العدد: 1
DOI: https://doi.org/10.22399/ijasrar.18
تاريخ النشر: 2025-03-07

الذكاء الاصطناعي في تحسين المالية وسلسلة التوريد: التحليلات التنبؤية لنمو الأعمال واستقرار السوق في الولايات المتحدة الأمريكية

تويوسي موتيلولا أولولا تيميلهين إيسايا أولاتوندي قسم الاتصالات، جامعة نورث داكوتا، غراند فوك، الولايات المتحدة الأمريكية*البريد الإلكتروني للمؤلف المراسل: motilola_olola56@yahoo.com قسم بناء بنية الشبكات التحتية، VEA، الاتصالات، مانشستر، المملكة المتحدةالبريد الإلكتروني: isaiah92_olatunde@gmail.com

تاريخ المقال:

DOI: 10.22399/ijasrar. 18
تاريخ الاستلام: 3 يناير 2025
تم القبول: 7 مارس 2025

الكلمات المفتاحية:

الذكاء الاصطناعي، تعلم الآلة، التحليلات التنبؤية، تحسين المالية، إدارة سلسلة التوريد، كشف الاحتيال، توقع الطلب.

الملخص

تدرس هذه الدراسة تطبيق الذكاء الاصطناعي (AI) وتعلم الآلة (ML) في تحسين عمليات سلسلة التوريد والتنبؤ المالي في الولايات المتحدة الأمريكية. تبحث البحث في كيفية تعزيز التحليلات التنبؤية المدفوعة بالذكاء الاصطناعي لنمو الأعمال واستقرار الأسواق. يتم استخدام مجموعة متنوعة من نماذج تعلم الآلة لمعالجة التحديات المختلفة: تُستخدم شبكات الذاكرة طويلة وقصيرة المدى (LSTM) للتنبؤ بالتسلسل في المجالات المالية والاقتصادية، بينما تدعم تقنيات الانحدار اللوجستي، والغابات العشوائية، وتقنيات التعزيز اكتشاف الاحتيال. بالإضافة إلى ذلك، يتم تطبيق المحولات التلقائية وخوارزميات غابة العزل لتحديد المعاملات المالية غير العادية، وتقوم نماذج ARIMA بتوقع ارتفاعات الطلب والموسمية. لتحسين اللوجستيات، يتم استخدام التعلم المعزز (شبكات Q العميقة) لتحسين تخطيط الطرق، وتقوم الشبكات العصبية بتوقع فترات إعادة التخزين المثلى بناءً على أنماط الطلب. يتم استخدام XGBoost لتقييم حساسية الأسعار لدى العملاء وتحسين استراتيجيات التسعير. يتم تقييم أداء نماذج التنبؤ باستخدام متوسط الخطأ التربيعي الجذري (RMSE) ومتوسط النسبة المطلقة للخطأ (MAPE). في المقابل، يتم قياس فعالية اكتشاف الاحتيال من خلال الدقة، والاسترجاع، ودرجة F1، والمساحة تحت المنحنى (AUC-ROC). يتم تقييم نماذج اللوجستيات من خلال إجمالي وقت التسليم، وتقليل التكاليف، وزيادة الكفاءة، بينما يتم التحقق من توقعات إعادة التخزين من خلال الدقة، ومتوسط الخطأ التربيعي (MSE)، ومعدلات دوران المخزون. يتم تقييم استراتيجيات التسعير بناءً على تأثير الإيرادات، ومقاييس المرونة، ومعدلات احتفاظ العملاء.

1. المقدمة

1.1 الخلفية

لقد أدى دمج الذكاء الاصطناعي (AI) في تحسين المالية وإدارة سلسلة التوريد إلى تحويل عمليات الأعمال بشكل كبير من خلال تمكين التحليلات التنبؤية لتحسين اتخاذ القرار، وتقليل المخاطر، وزيادة الكفاءة [1]. في الولايات المتحدة، تستخدم الشركات نماذج الذكاء الاصطناعي للتنبؤ بحركات أسعار الأسهم، واكتشاف المعاملات الاحتيالية، وتحسين مسارات اللوجستيات [2،3]. تتطلب الأسواق المالية، التي تتميز بتقلبات عالية، نماذج ذكاء اصطناعي قوية مثل الشبكات العصبية طويلة وقصيرة المدى (LSTM) وXGBoost للتنبؤ الدقيق بأسعار الأسهم وتحليل حساسية الأسعار [4]. وبالمثل، تعزز نماذج الذكاء الاصطناعي في إدارة سلسلة التوريد الكفاءة التشغيلية من خلال تقليل البصمة الكربونية وتحسين التنبؤ بالطلب من خلال التعلم المعزز ونماذج السلاسل الزمنية [5]. تتيح التطورات في تطبيقات الذكاء الاصطناعي للشركات التخفيف من الاضطرابات الاقتصادية وتحسين مرونة سلسلة التوريد [6،7،8].

1.2 أهمية هذا البحث

تعتبر هذه الأبحاث ضرورية في معالجة عدم الاستقرار المالي، وعدم كفاءة سلسلة التوريد، وعدم القدرة على التنبؤ في السوق من خلال استخدام تحليلات تنبؤية مدفوعة بالذكاء الاصطناعي [1]. يتأثر الاقتصاد الأمريكي بعوامل اقتصادية كبرى وجيوسياسية متنوعة، مما يتطلب من الشركات اعتماد نماذج تقييم المخاطر المعززة بالذكاء الاصطناعي للحفاظ على استقرار السوق [3]. كما زادت الأنشطة المالية الاحتيالية، مما يبرز الحاجة إلى الذكاء الاصطناعي في اكتشاف الجرائم المالية ورسم خرائط المعاملات غير المشروعة [2]. بالإضافة إلى ذلك، فإن نماذج تحسين سلسلة التوريد المدفوعة بالذكاء الاصطناعي، مثل الشبكات العصبية والتعزيز
التعلم، يوفر إدارة المخزون واللوجستيات في الوقت الفعلي، مما يقلل من الاختناقات التشغيلية ويحسن من فعالية التسليمات [9،10]. تستكشف هذه الدراسة كيف أن التحليلات التنبؤية لا تعزز فقط نمو الأعمال ولكنها تساهم أيضًا في الاستدامة الاقتصادية من خلال التخفيف من المخاطر وتحسين اتخاذ القرارات المالية.

1.3 الأهداف

الهدف الرئيسي من هذا البحث هو التحقيق في دور الذكاء الاصطناعي في تحسين الأداء المالي وإدارة سلسلة التوريد، مع التركيز على التحليلات التنبؤية لنمو الأعمال واستقرار السوق. أحد المحاور الرئيسية هو تطوير نماذج الذكاء الاصطناعي لتوقع أسعار الأسهم وتقييم المخاطر المالية، مما يمكّن الشركات من اتخاذ قرارات استثمارية باستخدام البيانات والتقليل من المخاطر المحتملة في السوق. كما يتناول هذا البحث فعالية نماذج كشف الاحتيال في تحديد المعاملات المالية غير العادية، مما يضمن نزاهة الأنظمة المالية من خلال تقليل الأنشطة الاحتيالية. هدف آخر حاسم هو تحسين عمليات سلسلة التوريد من خلال الاستفادة من الذكاء الاصطناعي في توقع الطلب، وإدارة المخزون، وتخطيط اللوجستيات، مما يحسن الكفاءة التشغيلية ويقلل التكاليف. يسعى البحث أيضًا إلى تقييم تأثير التحليلات التنبؤية على استقرار السوق ومرونة الأعمال، لا سيما في البيئات الاقتصادية المتقلبة حيث تكون اتخاذ القرارات الاستراتيجية أمرًا حاسمًا. أخيرًا، يهدف هذا البحث إلى استكشاف الفوائد البيئية والاقتصادية لتحسين سلسلة التوريد المدفوعة بالذكاء الاصطناعي، لا سيما في تقليل البصمة الكربونية وتعزيز ممارسات الأعمال المستدامة. من خلال هذه الأهداف، يقدم البحث تحليلًا شاملاً للإمكانات التحويلية للذكاء الاصطناعي في الإدارة المالية وإدارة سلسلة التوريد.

2. مراجعة الأدبيات

استكشفت العديد من الدراسات تطبيق الذكاء الاصطناعي في تحسين الأداء المالي وإدارة سلسلة التوريد. قام حسن وآخرون (2024) بدراسة كيفية تقليل التحليلات التنبؤية المدفوعة بالذكاء الاصطناعي للبصمات الكربونية وتعزيز كفاءة سلسلة التوريد [1]. وبالمثل، قام خان وآخرون (2024) بتحليل كيفية تأثير نماذج الذكاء الاصطناعي على التنبؤ المالي، وخاصة توقعات سوق الأسهم بناءً على المؤشرات الاقتصادية والأحداث الجيوسياسية [6]. في كشف الاحتيال، قام رحمن وآخرون (2024) برسم أنماط المعاملات غير المشروعة باستخدام التعلم الآلي وتحليل الشبكات لتحسين كشف الجرائم المالية [2]. قارن سيزان وآخرون (2023) بين نماذج التعلم الآلي المختلفة، بما في ذلك LSTM وXGBoost، لتوقع سوق الأسهم، مع تسليط الضوء على دقتها التنبؤية وآثارها المالية [4]. كما استكشفت الأبحاث الحديثة كيفية تطبيق الذكاء الاصطناعي في منع الاحتيال وتعزيز استقرار السوق [11،12].

2.2 الفجوات والتحديات

على الرغم من التقدم في تطبيقات الذكاء الاصطناعي، لا تزال هناك عدة تحديات قائمة. أشار رحمن وآخرون (2025) إلى صعوبة التنبؤ بدقة بحركات سوق الأسهم خلال الأزمات الاقتصادية بسبب سلوكيات السوق غير المتوقعة [3]. بالإضافة إلى ذلك، أكد رحمن وآخرون (2024) على قيود نماذج كشف الاحتيال في تحديد الجرائم المالية المعقدة التي تتضمن تقنيات غير مشروعة متطورة [2]. أشار حسن وآخرون (2025) إلى تحدي التنبؤ بالطلب في الوقت الحقيقي في إدارة سلسلة التوريد، حيث تتطلب ظروف السوق الديناميكية نماذج ذكاء اصطناعي قابلة للتكيف بشكل كبير [9]. علاوة على ذلك، لا تزال القضايا المتعلقة بجودة البيانات، وقابلية تفسير النماذج، والاعتبارات الأخلاقية في اتخاذ القرارات المدفوعة بالذكاء الاصطناعي غير محلولة [13]. يتطلب معالجة هذه التحديات تحسينات مستمرة في خوارزميات الذكاء الاصطناعي، وتقنيات تكامل البيانات المحسنة، وأطر أخلاقية لتنفيذ الذكاء الاصطناعي بشكل مسؤول.

3. المنهجية

3.1 جمع البيانات والمعالجة المسبقة

مصادر البيانات

تستخدم هذه الدراسة مجموعات بيانات متنوعة من مجالات متعددة، مما يضمن تحليلًا شاملاً لتحسين المالية وسلسلة التوريد. يتم الحصول على البيانات المالية من التقارير المالية المتاحة للجمهور، ومؤشرات سوق الأسهم، والملفات التنظيمية، مما يوفر رؤى تاريخية وفي الوقت الحقيقي حول الأسهم.
تحركات الأسعار والاتجاهات المالية. يتم الحصول على بيانات سلسلة التوريد من الشركات الرائدة في مجال التجزئة واللوجستيات، بما في ذلك السجلات المعاملات، مستويات المخزون، وعمليات اللوجستيات، مما يوفر مقاييس مفصلة لتوقع الطلب وتحسين المسارات. يتم اشتقاق بيانات كشف الاحتيال من المؤسسات المالية، وتتكون من سجلات المعاملات، تقارير الأنشطة المشبوهة، وتواريخ المعاملات للعملاء المجهولين. تم هيكلة هذه المجموعات البيانية لدعم النمذجة التنبؤية عبر سيناريوهات الأعمال المختلفة، لضمان أداء قوي للنموذج.

معالجة البيانات المسبقة

تعتبر معالجة البيانات خطوة حاسمة لضمان موثوقية ودقة النماذج المستخدمة في هذا البحث. تمر مجموعات البيانات الخام بعدة مراحل تحويل، بدءًا من تنظيف البيانات، حيث يتم التعامل مع القيم المفقودة باستخدام تقنيات التعويض، ويتم إزالة الإدخالات المكررة للحفاظ على سلامة البيانات. يتم ترميز المتغيرات الفئوية باستخدام ترميز واحد-ساخن وترميز التسمية لتسهيل توافق نماذج التعلم الآلي. كما يتم تطبيع البيانات العددية لتحسين تقارب النموذج ودقته. بالنسبة لبيانات السلاسل الزمنية، يتم تطبيق تقنيات التنعيم لإزالة الضوضاء (الشكل 1) نظرًا لأن سلسلة أسعار الأسهم الخام كانت تحتوي على تقلبات وضوضاء، مما يؤثر على دقة التنبؤ، ويتم إجراء هندسة الميزات لتوليد رؤى إضافية، مثل المتوسطات المتحركة ومؤشرات الاتجاه. يتم اكتشاف ومعالجة الشذوذ في المعاملات المالية باستخدام خوارزميات قائمة على العزل، مما يضمن التقاط الأنماط الاحتيالية بشكل فعال. من مجموعة البيانات، تم حقن قيمتين متطرفتين (واحدة مرتفعة جدًا عند 500 وواحدة منخفضة جدًا عند 5) (الشكل 2). يتم التعامل مع القيم الشاذة باستخدام القص (تحديد الحد عند النسبة المئوية 95) وWinsorization (استبدال القيم المتطرفة بالحدود الإحصائية).
الشكل 1. رسم بياني خطي يقارن أسعار الأسهم الخام بالقيم الملساء.
الشكل 2. مخطط الصندوق يعرض القيم المتطرفة في مجموعة البيانات.

3.2 تطوير النموذج

تستخدم هذه الدراسة مجموعة من نماذج التعلم الآلي والتعلم العميق، كل منها مُحسّن لمهام تحليل البيانات التنبؤية المحددة. بالنسبة للتنبؤات المالية، تُستخدم نماذج السلاسل الزمنية مثل ARIMA وشبكات الذاكرة طويلة وقصيرة المدى (LSTM) للتنبؤ باتجاهات أسعار الأسهم والتقلبات الاقتصادية. في كشف الاحتيال، تُستخدم نماذج قائمة على التصنيف مثل الانحدار اللوجستي، وغابة عشوائية، وXGBoost للتمييز بين المعاملات الشرعية والاحتيالية. يتم دمج نماذج التشفير التلقائي ونموذج غابة العزل لتحديد الشذوذ في بيانات المعاملات المالية. يستفيد تحسين سلسلة التوريد من نماذج التعلم المعزز لتخطيط الطرق، مما يضمن عمليات لوجستية فعالة من حيث التكلفة والكفاءة. يتم تدريب الشبكات العصبية للتنبؤ بفترات إعادة تخزين المخزون المثلى، مما يقلل من نقص المخزون وسيناريوهات الفائض. يتم تنفيذ هذه النماذج مع تقنيات ضبط المعلمات لتحقيق أداء مثالي عبر تطبيقات متنوعة.

3.3 إجراءات تدريب النموذج والتحقق منه

تخضع كل نموذج لتدريب صارم وتحقق لضمان فعاليته وموثوقيته. يتم تقسيم مجموعات البيانات إلى مجموعات تدريب، تحقق، واختبار، عادةً ما تتبع نسبة 70-15-15 لتحقيق توازن بين كفاءة التدريب والتعميم. يتم تدريب نماذج التعلم المراقب باستخدام مجموعات بيانات موسومة، حيث توجه قيم الحقيقة الأساسية تعلم النموذج. يتم ضبط المعلمات الفائقة باستخدام تقنيات البحث الشبكي والبحث العشوائي لتحسين المعلمات مثل معدلات التعلم، أعماق الأشجار، وتكوينات الطبقات في نماذج التعلم العميق. بالنسبة للنماذج المعتمدة على التعلم العميق، يتم تطبيق تنظيم الإسقاط للتخفيف من الإفراط في التكيف. يتم تدريب النماذج باستخدام انحدار التدرج المصغر لتحسين الكفاءة الحسابية وتحسين التقارب. يتم استخدام تقنيات التحقق المتقاطع مثل تحقق K-fold لضمان أداء قوي عبر توزيعات بيانات مختلفة. بعد التدريب، يتم تقييم تعميم النموذج باستخدام بيانات اختبار غير مرئية للتحقق من دقة التنبؤ.

3.4 مقاييس تقييم الأداء

يتم تقييم فعالية كل نموذج باستخدام مجموعة من مقاييس التقييم المصممة لتناسب التطبيق المحدد. بالنسبة لنماذج التنبؤ المالي، يتم استخدام الجذر التربيعي لمتوسط الخطأ التربيعي (RMSE) ومتوسط النسبة المئوية المطلقة للخطأ (MAPE) لقياس دقة التنبؤ. يتم تقييم نماذج كشف الاحتيال بناءً على الدقة، والاسترجاع، ودرجة F1، والمساحة تحت المنحنى – خاصية التشغيل الاستقبالي (AUC-ROC) لتحديد فعالية التصنيف. يتم تقييم نماذج التعلم المعزز لتحسين سلسلة التوريد باستخدام مقاييس مثل إجمالي وقت التسليم، وتقليل التكاليف، وزيادة الكفاءة. يتم التحقق من صحة الشبكات العصبية لإدارة المخزون بناءً على متوسط الخطأ التربيعي (MSE) ومعدلات دوران المخزون، لضمان إدارة المخزون بشكل مثالي. يتم تقييم نماذج XGBoost المستخدمة لتحسين استراتيجية التسعير باستخدام تأثير الإيرادات، ومقاييس المرونة، ومعدلات الاحتفاظ بالعملاء. توفر هذه المقاييس تقييمًا شاملًا لأداء النموذج، مما يضمن أن التحليلات التنبؤية المدفوعة بالذكاء الاصطناعي تساهم بشكل فعال في الاستقرار المالي وكفاءة سلسلة التوريد.
الشكل 3. يقارن هذا الرسم البياني العمودي أداء التنبؤ لنموذجين من السلاسل الزمنية (LSTM و ARIMA).

4. النتائج والمناقشة

4.1 أداء النماذج

المخطط الأول (الشكل 3) يقارن بصريًا أداء التنبؤ لنموذجين من السلاسل الزمنية LSTM و ARIMA – باستخدام الجذر التربيعي لمتوسط الخطأ التربيعي (RMSE) ومتوسط النسبة المئوية المطلقة للخطأ (MAPE). تشير القيم المنخفضة لـ RMSE و MAPE إلى دقة تنبؤ أفضل. في البيانات المستخدمة، يظهر نموذج LSTM قيم RMSE و MAPE أقل مقارنة بـ ARIMA، مما يشير إلى أنه قد يكون أفضل في التقاط الأنماط غير الخطية في تحركات أسعار الأسهم.
يوضح الرسم البياني العمودي المجمّع (الشكل 4) أداء نماذج كشف الاحتيال مثل الغابة العشوائية، والغابة المعزولة، والانحدار اللوجستي عبر عدة مقاييس: الدقة، والاسترجاع، ودرجة F1، وAUC-ROC. القيم الأعلى في هذه المقاييس (الأقرب إلى 1) تشير إلى أداء أفضل. تميل الغابة العشوائية إلى التفوق على الغابة المعزولة عبر معظم المقاييس، مما يُظهر قوتها في كشف المعاملات الاحتيالية. تساعد هذه التصويرات في تحديد أي نموذج يوازن بشكل أفضل بين الإيجابيات الكاذبة والسلبيات الكاذبة. يوضح الرسم البياني العمودي (الشكل 5) المقاييس الرئيسية للأداء لنموذج التعلم المعزز، الشبكات العصبية العميقة (DQN)، المستخدمة في تحسين اللوجستيات. تشمل المقاييس تقليل الوقت الإجمالي للتسليم (%)، وتقليل التكلفة (%)، وزيادة الكفاءة (%). تشير النسبة الأعلى في كل فئة إلى أداء أفضل. على سبيل المثال، يمكن أن تشير تقليص وقت التسليم إلى تحسينات تشغيلية كبيرة. تساعد هذه التصويرية في تقييم مدى تحسين النموذج لعمليات اللوجستيات تحت ظروف محاكاة. تقدم التصويرية (الشكل 6) تحليلًا مقارنًا لنموذجين من نماذج التعلم الآلي، الشبكات العصبية لإدارة المخزون وXGBoost لتحسين استراتيجية التسعير باستخدام الأعمدة المجمعة.
الشكل 4. مقارنة أداء نماذج كشف الاحتيال.
الشكل 5. مؤشرات الأداء الرئيسية لشبكات Q العميقة (DQN)
الرسوم البيانية. يقيم الرسم البياني الأول أداء إدارة المخزون استنادًا إلى ثلاثة مقاييس رئيسية: الدقة، ومتوسط الخطأ التربيعي (MSE)، ودوران المخزون. يشير نظام إدارة المخزون الفعال إلى معدلات دقة ودوران مخزون أعلى، مصحوبة بقيم MSE أقل، مما يدل على تقليل أخطاء التنبؤ. يركز الرسم البياني الثاني على تحسين استراتيجية التسعير ويفحص تأثير الإيرادات (نسبة الزيادة)، ومقاييس المرونة، ومعدلات احتفاظ العملاء (كنسبة مئوية). يتميز نموذج التسعير المحسن بشكل جيد بتأثير إيرادات أعلى واحتفاظ بالعملاء، إلى جانب مقاييس مرونة تضمن أن تعديلات الأسعار توازن بشكل فعال بين الطلب والربحية. تقدم هذه التصورات رؤى قيمة حول فعالية النموذج، مما يسمح للشركات بتحديد أي نهج يحقق كفاءة تشغيلية أفضل وتأثير في السوق. على سبيل المثال، إذا أظهر XGBoost زيادة كبيرة في تأثير الإيرادات مقارنة بالنماذج الأخرى، فقد يكون الخيار المفضل لاستراتيجيات التسعير الديناميكية.
الشكل 6. مقارنة أداء نماذج إدارة المخزون وتحسين استراتيجيات التسعير.

4.2 المناقشة والأعمال المستقبلية

تظهر النتائج التجريبية والتصورات المقدمة في هذه الدراسة الإمكانات الكبيرة لتحليلات التنبؤ المدعومة بالذكاء الاصطناعي في تحسين العمليات المالية وسلسلة التوريد. تشير التحليلات المقارنة إلى أن النماذج المتقدمة مثل LSTM للتنبؤ المالي، وغابة عشوائية للكشف عن الاحتيال، وتعلم التعزيز لتحسين اللوجستيات، تظهر أداءً متفوقًا في مجالاتها المعنية. هذه النماذج لا تحسن فقط دقة التنبؤ ولكنها تساهم أيضًا في تعزيز عمليات اتخاذ القرار التي تدفع نمو الأعمال واستقرار السوق. على الرغم من هذه النتائج الواعدة، لا تزال هناك عدة تحديات قائمة. القضايا المتعلقة بجودة البيانات، وقابلية تفسير النماذج، ودمج بيانات البث المباشر في الوقت الحقيقي تستمر في تقييد نشر هذه الحلول الذكية في بيئات الأعمال الديناميكية. على سبيل المثال، بينما تلتقط نماذج التعلم العميق أنماطًا معقدة في الأسواق المالية، فإن طبيعتها “الصندوق الأسود” غالبًا ما تعقد تفسير العمليات القرارية الأساسية. علاوة على ذلك، فإن تباين بيانات سلسلة التوريد يتطلب تطوير نماذج أكثر تكيفًا يمكنها الاستجابة بسرعة لتقلبات السوق. يجب أن تتناول الأبحاث المستقبلية هذه القيود من خلال استكشاف عدة مجالات رئيسية. أولاً، سيكون تعزيز قابلية تفسير النماذج من خلال تقنيات مثل SHAP (تفسيرات شابلي الإضافية) أو LIME (تفسيرات محلية قابلة للتفسير وغير مرتبطة بالنموذج) أمرًا حاسمًا لبناء ثقة أصحاب المصلحة. ثانيًا، يمكن أن يؤدي دمج تدفقات البيانات في الوقت الحقيقي ومجموعات البيانات متعددة الأنماط إلى تحسين متانة النماذج وقدرتها على التكيف. يمكن أن تركز الأبحاث أيضًا على تطوير نماذج هجينة تجمع بين مزايا التعلم العميق والأساليب الإحصائية التقليدية لالتقاط الاتجاهات غير الخطية والخطية بشكل أكثر فعالية. علاوة على ذلك، يمكن أن تؤدي المزيد من التحقيقات في تقنيات التعلم المعزز المتقدمة لوجستيات واستراتيجيات التسعير الديناميكي إلى تحقيق فوائد تشغيلية كبيرة. على سبيل المثال، قد يؤدي تحسين التخطيط للطرق بالاعتماد على المحاكاة مع التعلم المعزز إلى تحسين تخطيط الطرق وتقليل أوقات التسليم بشكل أكبر. بالإضافة إلى ذلك، يبقى استكشاف التأثير طويل الأمد للتسعير الديناميكي على الاحتفاظ بالعملاء ونمو الإيرادات مجالًا مهمًا للدراسات المستقبلية.
لقد أبرزت الأعمال الأخيرة دور تقنيات الذكاء الاصطناعي المتقدمة في التنبؤ المالي وتحسين سلسلة التوريد [10،14]. كما أظهرت التحليلات المقارنة أن الاحتيال القائم على التعلم العميق
يمكن أن تتفوق نماذج الكشف بشكل كبير على الأساليب التقليدية في السيناريوهات الواقعية [15]. وقد أكدت الدراسات حول استقرار السوق والتسعير الديناميكي على فوائد دمج التحليلات التنبؤية مع نماذج اقتصادية قوية. تدعم الأبحاث الناشئة حول إدارة المخزون وتحسين استراتيجيات التسعير الحاجة إلى نماذج توازن بين الكفاءة التشغيلية والاستجابة للسوق [11،12،16]. يتم استخدام الذكاء الاصطناعي في تطبيقات مختلفة وتم الإبلاغ عنها في الأدبيات [17-23].

5. الخاتمة

تسلط هذه الدراسة الضوء على الإمكانيات الواسعة للذكاء الاصطناعي (AI) وتعلم الآلة (ML) في تحسين التنبؤات المالية وتحسين عمليات سلسلة التوريد. يُظهر استخدام نماذج الذكاء الاصطناعي المتقدمة، بما في ذلك LSTM، وغابة عشوائية، وXGBoost، وتعلم التعزيز، فعاليتها في التنبؤ بأسعار الأسهم، واكتشاف المعاملات الاحتيالية، وتعزيز اللوجستيات، وتحسين التنبؤ بالطلب. تشير النتائج إلى أن التحليلات التنبؤية باستخدام الذكاء الاصطناعي تساهم بشكل كبير في تحسين اتخاذ القرارات التجارية، وزيادة الكفاءة التشغيلية، واستقرار ظروف السوق. تظهر نماذج الذكاء الاصطناعي المختلفة نقاط قوة في تطبيقات محددة. توفر LSTM وARIMA توقعات مالية موثوقة، بينما تتفوق غابة عشوائية وغابة العزل في اكتشاف الاحتيال. يعزز تعلم التعزيز اللوجستيات من خلال تحسين أوقات التسليم وإدارة التكاليف، وتدعم الشبكات العصبية إدارة المخزون من خلال التنبؤ بفترات إعادة التخزين المثلى. على الرغم من هذه التقدمات، لا تزال بعض التحديات قائمة. تشكل قضايا مثل جودة البيانات غير المتسقة، والطلبات الحسابية العالية، وتعقيد تفسير نماذج الذكاء الاصطناعي عوائق أمام التبني الأوسع. بالإضافة إلى ذلك، تتطلب تكامل البيانات في الوقت الحقيقي والاعتبارات الأخلاقية في نشر الذكاء الاصطناعي مزيدًا من البحث والتنقيح. يجب أن تستكشف الدراسات المستقبلية نماذج الذكاء الاصطناعي الهجينة التي تدمج التعلم العميق مع طرق التنبؤ الاقتصادية التقليدية لتحسين كل من الدقة والشفافية. علاوة على ذلك، سيساهم دمج البيانات المالية وسلسلة التوريد في الوقت الحقيقي، إلى جانب الاستفادة من تقنيات التعلم الانتقالي، في تعزيز القدرة على التكيف في المشهد الاقتصادي المتطور بسرعة.

بيانات المؤلفين:

  • الموافقة الأخلاقية: البحث الذي تم إجراؤه لا يتعلق باستخدام البشر أو الحيوانات.
  • تعارض المصالح: يعلن المؤلفون أنهم ليس لديهم أي مصالح مالية متنافسة معروفة أو علاقات شخصية قد تبدو أنها تؤثر على العمل المبلغ عنه في هذه الورقة.
  • إقرار: يعلن المؤلفون أنهم ليس لديهم أي شخص أو شركة ليعترفوا بهم.
  • مساهمات المؤلفين: يعلن المؤلفون أن لهم حقوق متساوية في هذه الورقة.
  • معلومات التمويل: يعلن المؤلفون أنه لا يوجد تمويل يجب الإقرار به.
  • بيان توافر البيانات: البيانات التي تدعم نتائج هذه الدراسة متاحة عند الطلب من المؤلف المراسل. البيانات غير متاحة للجمهور بسبب قيود الخصوصية أو الأخلاق.

References

[1]Hasan, M. R., Islam, M. Z., Sumon, M. F. I., Osiujjaman, M., Debnath, P., & Pant, L. (2024). Integrating Artificial Intelligence and Predictive Analytics in Supply Chain Management to Minimize Carbon Footprint and Enhance Business Growth in the USA. Journal of Business and Management Studies, 6(4), 195-212.
[2] Rahman, A., Debnath, P., Ahmed, A., Dalim, H. M., Karmakar, M., Sumon, M. F. I., & Khan, M. A. (2024). Machine learning and network analysis for financial crime detection: Mapping and identifying illicit transaction patterns in global black money transactions. Gulf Journal of Advance Business Research, 2(6), 250-272.
[3] Rahman, M. K., Dalim, H. M., Reza, S. A., Ahmed, A., Zeeshan, M. A. F., Jui, A. H., & Nayeem, M. B. (2025). Assessing the Effectiveness of Machine Learning Models in Predicting Stock Price Movements During Energy Crisis: Insights from Shell’s Market Dynamics. Journal of Business and Management Studies, 7(1), 44-61.
[4] Sizan, M. M. H., Das, B. C., Shawon, R. E. R., Rana, M. S., Al Montaser, M. A., Chouksey, A., & Pant, L. (2023). AI-Enhanced Stock Market Prediction: Evaluating Machine Learning Models for Financial Forecasting in the USA. Journal of Business and Management Studies, 5(4), 152-166.
[5] Smith, R., & Robinson, J. (2024). Machine Learning for Market Volatility and Economic Forecasting. Journal of Economic Computation, 13(1), 34-56.
[6] Khan, M. T., Akter, R., Dalim, H. M., Sayeed, A. A., Anonna, F. R., Mohaimin, M. R., & Karmakar, M. (2024). Predictive Modeling of US Stock Market and Commodities: Impact of Economic Indicators and Geopolitical Events Using Machine. Journal of Economics, Finance and Accounting Studies, 6(6), 17-33.
[7] Nguyen, H., & Tran, D. (2023). Real-Time Data Streams in AI-Driven Supply Chain Management. International Journal of Supply Chain Technology, 9(2), 98-120.
[8] Nguyen, T., & Robinson, P. (2023). Enhancing Predictive Accuracy in Business Analytics through AI. Journal of Artificial Intelligence Research, 15(1), 45-67.
[9] Hasan, M. R., Islam, M. R., & Rahman, M. A. (2025). Developing and implementing AI-driven models for demand forecasting in US supply chains: A comprehensive approach to enhancing predictive accuracy. Edelweiss Applied Science and Technology, 9(1), 1045-1068.
[10] Lee, J., & Park, S. (2023). The Role of Machine Learning in Fraud Prevention and Market Stability. Journal of Business Intelligence, 11(4), 85-101.
[11] Davis, R., & Zhao, L. (2024). Predictive Analytics in Market Stability: A Hybrid AI Approach. Journal of Economics and Data Science, 8(1), 134-150.
[12] Patel, A., & Lee, H. (2023). AI-Driven Insights for Supply Chain Resilience. Logistics and Business Analytics Journal, 7(3), 178-195.
[13]Akter, R., Nasiruddin, M., Anonna, F. R., Mohaimin, M. R., Nayeem, M. B., Ahmed, A., & Alam, S. (2023). Optimizing Online Sales Strategies in the USA Using Machine Learning: Insights from Consumer Behavior. Journal of Business and Management Studies, 5(4).
[14] Kumar, S., & Gupta, R. (2024). AI-Based Risk Assessment Models for Financial Institutions. Computational Finance Journal, 10(3), 56-72.
[15] Martinez, P., & Chen, W. (2024). AI-Powered Stock Market Predictions: A Comparative Analysis. Journal of Financial Data Science, 14(2), 56-78.
[16] Chen, Y., Li, X., & Zhao, H. (2024). AI and Blockchain Integration for Secure Financial Transactions. Journal of Financial Technology, 12(2), 78-94.
[17] Jafar Ismail, R., Samar Jaafar Ismael, Dr. Sara Raouf Muhamad Amin, Wassan Adnan Hashim, & Israa Tahseen Ali. (2024). Survey of Multiple Destination Route Discovery Protocols. International Journal of Computational and Experimental Science and Engineering, 10(3). https://doi.org/10.22399/ijcesen. 385
[18] S. Menaka, & V. Selvam. (2025). Bibliometric Analysis of Artificial Intelligence on Consumer Purchase Intention in E-Retailing. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 1007
[19] ZHANG, J. (2025). Artificial intelligence contributes to the creative transformation and innovative development of traditional Chinese culture. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 860
[20] M.K. Sarjas, & G. Velmurugan. (2025). Bibliometric Insight into Artificial Intelligence Application in Investment. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 864
[21] G. Prabaharan, S. Vidhya, T. Chithrakumar, K. Sika, & M.Balakrishnan. (2025). AI-Driven Computational Frameworks: Advancing Edge Intelligence and Smart Systems. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 1165
[22] Serap ÇATLI DİNÇ, AKMANSU, M., BORA, H., ÜÇGÜL, A., ÇETİN, B. E., ERPOLAT, P., … ŞENTÜRK, E. (2024). Evaluation of a Clinical Acceptability of Deep Learning-Based Autocontouring: An Example of The Use of Artificial Intelligence in Prostate Radiotherapy. International Journal of Computational and Experimental Science and Engineering, 10(4). https://doi.org/10.22399/ijcesen. 386
[23] S. Esakkiammal, & K. Kasturi. (2024). Advancing Educational Outcomes with Artificial Intelligence: Challenges, Opportunities, And Future Directions. International Journal of Computational and Experimental Science and Engineering, 10(4). https://doi.org/10.22399/ijcesen. 799

Journal: International Journal of Applied Sciences and Radiation Research, Volume: 2, Issue: 1
DOI: https://doi.org/10.22399/ijasrar.18
Publication Date: 2025-03-07

Artificial Intelligence in Financial and Supply Chain Optimization: Predictive Analytics for Business Growth and Market Stability in the USA

Toyosi Motilola Olola , , Timilehin Isaiah Olatunde Department of Communications, University of North Dakota, Grand Folks, USA*Corresponding Author Email: motilola_olola56@yahoo.com Department of Network Infrastructure Building, VEA, Telecoms, Manchester, United KingdomEmail: isaiah92_olatunde@gmail.com

Article History:

DOI: 10.22399/ijasrar. 18
Received: Jan. 03, 2025
Accepted: Mar. 07, 2025

Keywords:

Artificial Intelligence, Machine Learning, Predictive Analytics, Financial Optimization, Supply Chain Management, Fraud Detection, Demand Forecasting.

Abstract

This study investigates the application of Artificial Intelligence (AI) and Machine Learning (ML) in optimizing supply chain operations and financial forecasting in the USA. The research examines how AI-driven predictive analytics can foster business growth and stabilize markets. A diverse set of ML models is employed to address various challenges: Long Short-Term Memory (LSTM) networks are used for sequence forecasting in financial and economic domains, while Logistic Regression, Random Forest, and Boosting techniques support fraud detection. Additionally, autoencoders and Isolation Forest algorithms are applied to identify unusual financial transactions, and ARIMA models forecast demand spikes and seasonality. For logistics optimization, Reinforcement Learning ( Deep Q-Networks) is used to improve route planning, and Neural Networks predict optimal restocking periods based on demand patterns. XGBoost is used to assess customer price sensitivity and optimize pricing strategies. The performance of forecasting models is evaluated using Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). In contrast, fraud detection effectiveness is measured through Precision, Recall, F1-score, and the Area Under the Curve (AUC-ROC). Logistics models are assessed by Total Delivery Time, Cost Reduction, and Efficiency Gains while restocking predictions are validated via accuracy, Mean Squared Error (MSE), and inventory turnover rates. Pricing strategies are evaluated based on Revenue Impact, Elasticity Metrics, and Customer Retention Rates.

1. Introduction

1.1 Background

Integrating Artificial Intelligence (AI) in financial optimization and supply chain management has significantly transformed business operations by enabling predictive analytics to improve decisionmaking, minimize risks, and enhance efficiency [1]. In the United States, businesses are making use of AI models to predict stock price movements, detect fraudulent transactions, and optimize logistics routes [2,3]. Financial markets, characterized by high volatility, require robust AI models such as Long ShortTerm Memory (LSTM) networks and XGBoost for accurate stock price forecasting and price sensitivity analysis [4]. Similarly, AI models in supply chain management enhance operational efficiency by reducing carbon footprints and optimizing demand forecasting through Reinforcement Learning and time series models [5]. The advancements in AI applications allow businesses to mitigate economic disruptions and improve supply chain resilience [6,7,8].

1.2 Importance Of This Research

This research is vital in addressing financial instability, supply chain inefficiencies, and market unpredictability by utilizing AI-driven predictive analytics [1]. The U.S. economy is influenced by various macroeconomic and geopolitical factors, requiring businesses to adopt AI-enhanced risk assessment models to sustain market stability [3]. Fraudulent financial activities have also surged, emphasizing the need for AI in financial crime detection and illicit transaction mapping [2]. Additionally, AI-driven supply chain optimization models, such as Neural Networks and Reinforcement
Learning, provide real-time inventory and logistics management, this reduces operational bottlenecks and improve the efficacy of deliveries [9,10]. This research explores how predictive analytics not only enhances business growth but also contributes to economic sustainability by mitigating risks and optimizing financial decision-making.

1.3 Objectives

The primary objective of this research is to investigate the role of AI in financial optimization and supply chain management, focusing on predictive analytics for business growth and market stability. One key focus is the development of AI models for stock price prediction and financial risk assessment, enabling businesses to make investment decisions using data and mitigate potential market risks. This study also examines the effectiveness of fraud detection models in identifying unusual financial transactions, which ensures the integrity of financial systems by minimizing fraudulent activities. Another critical objective is optimizing supply chain operations by leveraging AI for demand forecasting, inventory management, and logistics planning, thereby improving operational efficiency and reducing costs. The research also seeks to assess the impact of predictive analytics on market stability and business resilience, particularly in volatile economic environments where strategic decision-making is crucial. Lastly, this study aims to explore the environmental and economic benefits of AI-driven supply chain optimization, particularly in reducing carbon footprints and promoting sustainable business practices. Through these objectives, the research provides a comprehensive analysis of AI’s transformative potential in financial and supply chain management.

2. Literature Review

Numerous studies have explored the application of AI in financial optimization and supply chain management. Hasan et al. (2024) examined how AI-driven predictive analytics minimize carbon footprints and enhance supply chain efficiency [1]. Similarly, Khan et al. (2024) analyzed how AI models impact financial forecasting, particularly stock market predictions based on economic indicators and geopolitical events [6]. In fraud detection, Rahman et al. (2024) mapped illicit transaction patterns using machine learning and network analysis to improve financial crime detection [2]. Sizan et al. (2023) compared various machine learning models, including LSTM and XGBoost, for stock market prediction, highlighting their predictive accuracy and financial implications [4]. Recent research has also explored how AI can be applied in fraud prevention and promotion of market stability [11,12].

2.2 Gaps and Challenges

Despite the advancements in AI applications, several challenges persist. Rahman et al. (2025) highlighted the difficulty of accurately predicting stock market movements during economic crises due to unpredictable market behaviors [3]. Additionally, Rahman et al. (2024) emphasized the limitations of fraud detection models in identifying sophisticated financial crimes involving evolving illicit techniques [2]. Hasan et al. (2025) pointed out the challenge of real-time demand forecasting in supply chain management, where dynamic market conditions require highly adaptive AI models [9]. Furthermore, issues related to data quality, model interpretability, and ethical considerations in AI-driven decisionmaking remain unresolved [13]. Addressing these challenges requires continuous improvements in AI algorithms, enhanced data integration techniques, and ethical frameworks for responsible AI implementation.

3. Methodology

3.1 Data Collection and Preprocessing

Data Sources

This study utilizes diverse datasets from multiple domains, ensuring a comprehensive analysis of financial and supply chain optimization. Financial data is sourced from publicly available financial reports, stock market indices, and regulatory filings, offering real-time and historical insights into stock
price movements and financial trends. Supply chain data is obtained from leading retail and logistics companies, including transactional records, inventory levels, and logistics operations, providing detailed metrics for demand forecasting and route optimization. Fraud detection data is derived from financial institutions, comprising transactional logs, suspicious activity reports, and anonymized customer transaction histories. These datasets are structured to support predictive modeling across various business scenarios, to ensure robust model performance.

Data Preprocessing

Data preprocessing is a critical step in ensuring the reliability and accuracy of the models used in this research. The raw datasets undergo multiple transformation stages, starting with data cleaning, where missing values are handled using imputation techniques, and duplicate entries are removed to maintain data integrity. Categorical variables are encoded using one-hot encoding and label encoding to facilitate machine learning model compatibility. Numerical data is also normalized to improve model convergence and accuracy. For time-series data, smoothing techniques are applied to remove noise (Figure 1) since the raw stock price series had fluctuations and noise, affecting forecasting accuracy., and feature engineering is conducted to generate additional insights, such as moving averages and trend indicators. Anomalies in financial transactions are detected and treated using isolation-based algorithms, ensuring that fraudulent patterns are effectively captured. From the dataset, two extreme values were injected (one very high at 500 and one very low at 5 ) (Figure 2). Outliers are handled using clipping (capping at the 95th percentile) and Winsorization (replacing extreme values with statistical limits).
Figure 1. Line plot comparing raw stock prices with smoothed values.
Figure 2. A boxplot displaying extreme values in the dataset.

3.2 Model Development

This study employs a range of machine learning and deep learning models, each optimized for specific predictive analytics tasks. For financial forecasting, time-series models such as ARIMA and Long ShortTerm Memory (LSTM) networks are used to predict stock price trends and economic fluctuations. In fraud detection, classification-based models such as Logistic Regression, Random Forest, and XGBoost are employed to distinguish between legitimate and fraudulent transactions. Autoencoders and Isolation Forest models are integrated to identify anomalies in financial transaction data. Supply chain optimization leverages reinforcement learning models for route planning, ensuring cost-effective and efficient logistics operations. Neural networks are trained to predict optimal inventory restocking periods, minimizing stock shortages and overstock scenarios. These models are implemented with hyperparameter tuning techniques to achieve optimal performance across various applications.

3.3 Model Training and Validation Procedures

Each model undergoes rigorous training and validation to ensure its effectiveness and reliability. The datasets are split into training, validation, and testing sets, typically following a 70-15-15 ratio to balance training efficiency and generalization. Supervised learning models are trained using labeled datasets, where ground truth values guide model learning. Hyperparameter tuning is conducted using grid search and random search techniques to optimize parameters such as learning rates, tree depths, and layer configurations in deep learning models. For deep learning-based models, dropout regularization is applied to mitigate overfitting. The models are trained using mini-batch gradient descent to optimize computational efficiency and improve convergence. Cross-validation techniques such as K-fold validation are employed to ensure robust performance across different data distributions. After training, model generalization is assessed using unseen test data to validate predictive accuracy.

3.4 Performance Evaluation Metrics

The effectiveness of each model is assessed using a range of evaluation metrics tailored to the specific application. For financial forecasting models, Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE) are used to measure prediction accuracy. Fraud detection models are evaluated based on Precision, Recall, F1-score, and the Area Under the Curve – Receiver Operating Characteristic (AUC-ROC) to determine classification effectiveness. Reinforcement learning models for supply chain optimization are assessed using metrics such as Total Delivery Time, Cost Reduction, and Efficiency Gains. Neural networks for inventory management are validated based on Mean Squared Error (MSE) and inventory turnover rates, ensuring optimal stock management. XGBoost models used for pricing strategy optimization are evaluated using Revenue Impact, Elasticity Metrics, and Customer Retention Rates. These metrics provide a holistic assessment of model performance, ensuring that AIdriven predictive analytics contribute effectively to financial stability and supply chain efficiency.
Figure 3. This bar chart compares the forecasting performance of two time-series models(LSTM and ARIMA).

4. Results and Discussion

4.1 Model Performances

The first chart (Figure 3) visually compares the forecasting performance of two time-series models LSTM and ARIMA – using the Root Mean Squared Error (RMSE) and Mean Absolute Percentage Error (MAPE). Lower RMSE and MAPE values indicate better forecasting accuracy. In the data used, the LSTM model shows lower RMSE and MAPE values compared to ARIMA, indicating that it might be better at capturing the nonlinear patterns in stock price movements.
The grouped bar (Figure 4) chart compares the performance of fraud detection models such as Random Forest, Isolation Forest, and Logistic Regression across multiple metrics: Precision, Recall, F1-score, and AUC-ROC. Higher values in these metrics (closer to 1) indicate better performance. Random Forest tends to outperform Isolation Forest across most metrics, demonstrating its robustness in detecting fraudulent transactions. This visualization helps in identifying which model best balances the trade-offs between false positives and false negatives. The bar chart (Figure 5) illustrates key performance metrics for a reinforcement learning model, Deep Q-Networks(DQN), used in logistics optimization. The metrics include Total Delivery Time Reduction (%), Cost Reduction (%), and Efficiency Gains (%). A higher percentage in each category indicates better performance. For instance, a reduction in delivery time could imply significant operational improvements. This visualization assists in evaluating how well the model improves logistics operations under simulated conditions. The visualization (Figure 6) presents a comparative analysis of two machine learning models, Neural Networks for inventory management and XGBoost for pricing strategy optimization using grouped bar
Figure 4. Performance comparison of fraud detection models.
Figure 5. Key performance metrics for Deep Q-Networks(DQN)
charts. The first chart evaluates inventory management performance based on three key metrics: Accuracy, Mean Squared Error (MSE), and Inventory Turnover. An effective stock management system is indicated by higher accuracy and inventory turnover rates, coupled with lower MSE values, signifying reduced forecasting errors. The second chart focuses on pricing strategy optimization and examines Revenue Impact (percentage increase), Elasticity Metrics, and Customer Retention Rates (percentage). A well-optimized pricing model is characterized by higher revenue impact and customer retention, along with elasticity metrics that ensure pricing adjustments effectively balance demand and profitability. These visualizations offer valuable insights into model effectiveness, allowing businesses to determine which approach yields better operational efficiency and market influence. For example, if XGBoost demonstrates a substantial increase in revenue impact compared to other models, it may be the preferred choice for dynamic pricing strategies.
Figure 6. Performance comparison of inventory management and pricing strategy optimization models.

4.2 Discussion and Future Works

The experimental results and visualizations presented in this study demonstrate the significant potential of AI-powered predictive analytics in optimizing financial and supply chain operations. The comparative analysis indicates that advanced models such as LSTM for financial forecasting, Random Forest for fraud detection, and reinforcement learning for logistics optimization, exhibit superior performance in their respective domains . These models not only improve predictive accuracy but also contribute to enhanced decision-making processes that drive business growth and market stability. Despite these promising findings, several challenges remain. Issues related to data quality, model interpretability, and the integration of real-time streaming data continue to constrain the deployment of these AI solutions in dynamic business environments. For instance, while deep learning models capture complex patterns in financial markets, their “black-box” nature often complicates the explanation of underlying decision processes. Furthermore, the variability of supply chain data necessitates the development of more adaptive models that can respond swiftly to market fluctuations. Future research should address these limitations by exploring several key avenues. First, enhancing model interpretability through techniques such as SHAP (Shapley Additive Explanations) or LIME (Local Interpretable Model-agnostic Explanations) will be critical to building stakeholder trust. Second, incorporating real-time data streams and multi-modal datasets can further improve model robustness and adaptability. Research could also focus on developing hybrid models that combine the strengths of deep learning with traditional statistical approaches to capture both nonlinear and linear trends more effectively. Moreover, further investigations into advanced reinforcement learning techniques for logistics and dynamic pricing strategies could yield significant operational benefits. For example, simulation-based optimization combined with reinforcement learning may enhance route planning and reduce delivery times even further. Additionally, exploring the long-term impact of dynamic pricing on customer retention and revenue growth remains an important area for future studies.
Recent works have highlighted the role of advanced AI techniques in financial forecasting and supply chain optimization [10,14]. Comparative analyses have also shown that deep learning-based fraud
detection models can substantially outperform traditional approaches in real-world scenarios [15]. Studies on market stability and dynamic pricing have emphasized the benefits of integrating predictive analytics with robust economic models. Emerging research on inventory management and pricing strategy optimization further supports the need for models that balance operational efficiency with market responsiveness [11,12,16]. Artificial Intelligence is used in different application and reported in the literatüre [17-23].

5. Conclusion

This study highlights the extensive potential of artificial intelligence (AI) and machine learning (ML) in improving financial forecasting and optimizing supply chain operations. The use of advanced AI models, including LSTM, Random Forest, XGBoost, and Reinforcement Learning, demonstrates their effectiveness in predicting stock prices, detecting fraudulent transactions, enhancing logistics, and refining demand forecasting. The findings indicate that predictive analytics using Artificial Intelligence contribute significantly to improving business decision-making, increasing operational efficiency, and stabilizing market conditions. Different AI models exhibit strengths in specific applications. LSTM and ARIMA provide reliable financial forecasts, while Random Forest and Isolation Forest excel in fraud detection. Reinforcement Learning enhances logistics by optimizing delivery times and cost management, and Neural Networks support inventory management by forecasting optimal restocking periods. Despite these advancements, some challenges remain. Issues such as inconsistent data quality, high computational demands, and the complexity of AI model interpretation pose barriers to broader adoption. Additionally, real-time data integration and ethical considerations in AI deployment require further research and refinement. Future studies should explore hybrid AI models that integrate deep learning with traditional econometric forecasting methods to improve both accuracy and transparency. Moreover, incorporating real-time financial and supply chain data, along with leveraging transfer learning techniques, will enhance adaptability in rapidly evolving economic landscapes.

Author Statements:

  • Ethical approval: The conducted research is not related to either human or animal use.
  • Conflict of interest: The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper
  • Acknowledgement: The authors declare that they have nobody or no-company to acknowledge.
  • Author contributions: The authors declare that they have equal right on this paper.
  • Funding information: The authors declare that there is no funding to be acknowledged.
  • Data availability statement: The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

References

[1]Hasan, M. R., Islam, M. Z., Sumon, M. F. I., Osiujjaman, M., Debnath, P., & Pant, L. (2024). Integrating Artificial Intelligence and Predictive Analytics in Supply Chain Management to Minimize Carbon Footprint and Enhance Business Growth in the USA. Journal of Business and Management Studies, 6(4), 195-212.
[2] Rahman, A., Debnath, P., Ahmed, A., Dalim, H. M., Karmakar, M., Sumon, M. F. I., & Khan, M. A. (2024). Machine learning and network analysis for financial crime detection: Mapping and identifying illicit transaction patterns in global black money transactions. Gulf Journal of Advance Business Research, 2(6), 250-272.
[3] Rahman, M. K., Dalim, H. M., Reza, S. A., Ahmed, A., Zeeshan, M. A. F., Jui, A. H., & Nayeem, M. B. (2025). Assessing the Effectiveness of Machine Learning Models in Predicting Stock Price Movements During Energy Crisis: Insights from Shell’s Market Dynamics. Journal of Business and Management Studies, 7(1), 44-61.
[4] Sizan, M. M. H., Das, B. C., Shawon, R. E. R., Rana, M. S., Al Montaser, M. A., Chouksey, A., & Pant, L. (2023). AI-Enhanced Stock Market Prediction: Evaluating Machine Learning Models for Financial Forecasting in the USA. Journal of Business and Management Studies, 5(4), 152-166.
[5] Smith, R., & Robinson, J. (2024). Machine Learning for Market Volatility and Economic Forecasting. Journal of Economic Computation, 13(1), 34-56.
[6] Khan, M. T., Akter, R., Dalim, H. M., Sayeed, A. A., Anonna, F. R., Mohaimin, M. R., & Karmakar, M. (2024). Predictive Modeling of US Stock Market and Commodities: Impact of Economic Indicators and Geopolitical Events Using Machine. Journal of Economics, Finance and Accounting Studies, 6(6), 17-33.
[7] Nguyen, H., & Tran, D. (2023). Real-Time Data Streams in AI-Driven Supply Chain Management. International Journal of Supply Chain Technology, 9(2), 98-120.
[8] Nguyen, T., & Robinson, P. (2023). Enhancing Predictive Accuracy in Business Analytics through AI. Journal of Artificial Intelligence Research, 15(1), 45-67.
[9] Hasan, M. R., Islam, M. R., & Rahman, M. A. (2025). Developing and implementing AI-driven models for demand forecasting in US supply chains: A comprehensive approach to enhancing predictive accuracy. Edelweiss Applied Science and Technology, 9(1), 1045-1068.
[10] Lee, J., & Park, S. (2023). The Role of Machine Learning in Fraud Prevention and Market Stability. Journal of Business Intelligence, 11(4), 85-101.
[11] Davis, R., & Zhao, L. (2024). Predictive Analytics in Market Stability: A Hybrid AI Approach. Journal of Economics and Data Science, 8(1), 134-150.
[12] Patel, A., & Lee, H. (2023). AI-Driven Insights for Supply Chain Resilience. Logistics and Business Analytics Journal, 7(3), 178-195.
[13]Akter, R., Nasiruddin, M., Anonna, F. R., Mohaimin, M. R., Nayeem, M. B., Ahmed, A., & Alam, S. (2023). Optimizing Online Sales Strategies in the USA Using Machine Learning: Insights from Consumer Behavior. Journal of Business and Management Studies, 5(4).
[14] Kumar, S., & Gupta, R. (2024). AI-Based Risk Assessment Models for Financial Institutions. Computational Finance Journal, 10(3), 56-72.
[15] Martinez, P., & Chen, W. (2024). AI-Powered Stock Market Predictions: A Comparative Analysis. Journal of Financial Data Science, 14(2), 56-78.
[16] Chen, Y., Li, X., & Zhao, H. (2024). AI and Blockchain Integration for Secure Financial Transactions. Journal of Financial Technology, 12(2), 78-94.
[17] Jafar Ismail, R., Samar Jaafar Ismael, Dr. Sara Raouf Muhamad Amin, Wassan Adnan Hashim, & Israa Tahseen Ali. (2024). Survey of Multiple Destination Route Discovery Protocols. International Journal of Computational and Experimental Science and Engineering, 10(3). https://doi.org/10.22399/ijcesen. 385
[18] S. Menaka, & V. Selvam. (2025). Bibliometric Analysis of Artificial Intelligence on Consumer Purchase Intention in E-Retailing. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 1007
[19] ZHANG, J. (2025). Artificial intelligence contributes to the creative transformation and innovative development of traditional Chinese culture. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 860
[20] M.K. Sarjas, & G. Velmurugan. (2025). Bibliometric Insight into Artificial Intelligence Application in Investment. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 864
[21] G. Prabaharan, S. Vidhya, T. Chithrakumar, K. Sika, & M.Balakrishnan. (2025). AI-Driven Computational Frameworks: Advancing Edge Intelligence and Smart Systems. International Journal of Computational and Experimental Science and Engineering, 11(1). https://doi.org/10.22399/ijcesen. 1165
[22] Serap ÇATLI DİNÇ, AKMANSU, M., BORA, H., ÜÇGÜL, A., ÇETİN, B. E., ERPOLAT, P., … ŞENTÜRK, E. (2024). Evaluation of a Clinical Acceptability of Deep Learning-Based Autocontouring: An Example of The Use of Artificial Intelligence in Prostate Radiotherapy. International Journal of Computational and Experimental Science and Engineering, 10(4). https://doi.org/10.22399/ijcesen. 386
[23] S. Esakkiammal, & K. Kasturi. (2024). Advancing Educational Outcomes with Artificial Intelligence: Challenges, Opportunities, And Future Directions. International Journal of Computational and Experimental Science and Engineering, 10(4). https://doi.org/10.22399/ijcesen. 799