Geri Dön

A comparative study of machine learning classification alorithms on acceleration data

Hızlanma verileri üzerinde makine eğitimi sınıflandırma aloritmalarının karşılaştırmalı bir çalışması

  1. Tez No: 840875
  2. Yazar: MUHAMMAD AHMED MAJEED
  3. Danışmanlar: Assist. Prof. Dr. NURİ BİNGÖL, DR. ÖĞR. ÜYESİ FAYYAZ AHMAD
  4. Tez Türü: Yüksek Lisans
  5. Konular: Bilgisayar Mühendisliği Bilimleri-Bilgisayar ve Kontrol, Computer Engineering and Computer Science and Control
  6. Anahtar Kelimeler: Belirtilmemiş.
  7. Yıl: 2023
  8. Dil: İngilizce
  9. Üniversite: Üsküdar Üniversitesi
  10. Enstitü: Fen Bilimleri Enstitüsü
  11. Ana Bilim Dalı: Bilgisayar Mühendisliği Ana Bilim Dalı
  12. Bilim Dalı: Bilgisayar Mühendisliği Bilim Dalı
  13. Sayfa Sayısı: Belirtilmemiş.

Özet

Human activity recognition using machine learning has gained significant attention in recent years. The application of machine learning techniques to recognize human activities by analyzing acceleration data is focused on in this thesis research. The motivation for this research arises from the increasing popularity of wristband gadgets, mobile applications, smartwatches, etc., which collect gyroscope, gravity, rotation, and acceleration data, as well as the advancements in machine learning. Python and MATLAB® code was used in this research to extract 66 features from raw acceleration data associated with six human activities: walking, standing, sitting, upstairs, downstairs, and jogging. Four classification algorithms were implemented: Decision Tree, Support Vector Machine (SVM), Naïve Bayes, and K Nearest Neighbors (KNN). Classification models capable of accurately predicting various human activities were created using these algorithms. Across the different algorithms, the accuracy levels of the classification models varied. An accuracy of 87.4% was achieved by the Decision Tree algorithm, 90.9% by Naïve Bayes, 95.3% by Support Vector Machine, and 88.0% by KNN. Several metrics, including accuracy, precision, recall, and F1-score, were used to evaluate the performance of each algorithm. The numerical experimental results demonstrate that each algorithm has its strengths and limitations in recognizing human activities. Among these algorithms, the Support Vector Machine algorithm demonstrated the highest accuracy of 95.3%. The effectiveness of machine learning techniques in human activity recognition is showcased in this study, highlighting the superiority of the Support Vector Machine algorithm in accurately identifying human activities using raw data.

Özet (Çeviri)

Human activity recognition using machine learning has gained significant attention in recent years. The application of machine learning techniques to recognize human activities by analyzing acceleration data is focused on in this thesis research. The motivation for this research arises from the increasing popularity of wristband gadgets, mobile applications, smartwatches, etc., which collect gyroscope, gravity, rotation, and acceleration data, as well as the advancements in machine learning. Python and MATLAB® code was used in this research to extract 66 features from raw acceleration data associated with six human activities: walking, standing, sitting, upstairs, downstairs, and jogging. Four classification algorithms were implemented: Decision Tree, Support Vector Machine (SVM), Naïve Bayes, and K Nearest Neighbors (KNN). Classification models capable of accurately predicting various human activities were created using these algorithms. Across the different algorithms, the accuracy levels of the classification models varied. An accuracy of 87.4% was achieved by the Decision Tree algorithm, 90.9% by Naïve Bayes, 95.3% by Support Vector Machine, and 88.0% by KNN. Several metrics, including accuracy, precision, recall, and F1-score, were used to evaluate the performance of each algorithm. The numerical experimental results demonstrate that each algorithm has its strengths and limitations in recognizing human activities. Among these algorithms, the Support Vector Machine algorithm demonstrated the highest accuracy of 95.3%. The effectiveness of machine learning techniques in human activity recognition is showcased in this study, highlighting the superiority of the Support Vector Machine algorithm in accurately identifying human activities using raw data.

Benzer Tezler

  1. Onkolojik amaçlı pozitron emisyon tomografisi/bilgisayarlı tomografi yapılan hastalarda, makine öğrenimi yöntemiyle malign/benign lezyonların ayırıcı tanısı

    Differential diagnosis of malign/benign lesions using machine learning method in patients who had positron emission tomography/computed tomography for oncological purposes

    AYKUT HASBEK

    Yüksek Lisans

    Türkçe

    Türkçe

    2024

    BiyomühendislikSivas Cumhuriyet Üniversitesi

    Biyomühendislik Ana Bilim Dalı

    PROF. DR. MEHTAP ERŞAN

    DR. ÖĞR. ÜYESİ EMRE GÖRGÜN

  2. Ovulasyon indüksiyonu tedavisinde folliküler gelişimin ultrasonografik takibi

    Başlık çevirisi yok

    MERİH BAYRAM

    Tıpta Uzmanlık

    Türkçe

    Türkçe

    1987

    Kadın Hastalıkları ve DoğumGazi Üniversitesi

    Kadın Hastalıkları ve Doğum Ana Bilim Dalı

    DOÇ. DR. MÜLAZIM YILDIRIM

  3. Çimentonun sertleşmesi üzerinde kimyasal komponentlerin etkisi

    Başlık çevirisi yok

    NACİYE TÜRKEL

    Yüksek Lisans

    Türkçe

    Türkçe

    1986

    Kimya MühendisliğiUludağ Üniversitesi

    Kimya Ana Bilim Dalı

    PROF. DR. MUSTAFA CEBE

  4. Bazı esterlerin kinetik incelenmesi ve termodinamik parametrelerin belirlenmesi

    Kinetic study and detesmination of thermodnamic parameters of some esters

    İBRAHİM TAŞ

    Yüksek Lisans

    Türkçe

    Türkçe

    1986

    Kimya MühendisliğiUludağ Üniversitesi

    Kimya Ana Bilim Dalı

    PROF. DR. MUSTAFA CEBE