?url_ver=Z39.88-2004&rft_id=1617051113&rft_val_fmt=info%3Aofi%2Ffmt%3Akev%3Amtx%3Adc&rft.title=PENDEKATAN+OVERSAMPLING+SMOTE+UNTUK%0D%0AIMBALANCED+DATASET+AKSARA+LAMPUNG+DAN%0D%0AKLASIFIKASI+MENGGUNAKAN+SVM&rft.creator=EDO+PRIYONO+MUALIM%2C+1617051113&rft.subject=000+Ilmu+komputer%2C+informasi+dan+pekerjaan+umum&rft.subject=500+ilmu+pengetahuan+alam+dan+matematika&rft.description=Klasifikasi+gambar+adalah+pembelajaran+supervised+dengan+mendifinisikan+kelas+gambar%2C+kemudian+melatih+sebuah+model+menggunakan+data+training%2C+validasi%2C+dan+testing+untuk+memprediksi+label+gambar+tulisan+tangan+Aksara+Lampung.+Tujuan+penelitian+ini+adalah+untuk+mengembangkan+model+Support+Vector+Machine+(SVM)+yang+mampu+mengklasifikasi+kelas+gambar+secara+akurat.+Teknik+klasifikasi+SVM+dipilih+karena+dapat+memproses+data+dalam+bentuk+tulisan+dan+gambar%2C+memiliki+implementasi+kernel+sehingga+memungkinkan+data+yang+tidak+dapat+dibagi+secara+linear+untuk+dikonversikan+ke+data+dimensi+lebih+tinggi.+Hasil+penelitian+dapat+digunakan+sebagai+panduan+untuk+penelitian+selanjutnya+dengan+fokus+topik+yang+lebih+mendalam%2C+terutama+untuk+subjek+mengenai+klasifikasi+gambar.+Penelitian+ini+menggunakan+karakter+tradisional+dari+Kota+Lampung+di+Indonesia.+Berbeda+dengan+alfabet+latin+umumnya%2C+Bahasa+Lampung+memiliki+20+huruf+induk%2C+12+anak+huruf%2C+dan+tanda+baca+khusus.+Dataset+Lampung+yang+digunakan+terdiri+dari+18+huruf+induk+dan+tersebar+di+32140+gambar+grayscale+tanpa+anak+huruf+dan+tanda+baca.+Akan+tetapi+data+tidak+memiliki+pesebaran+data+yang+seimbang.+Penelitian+ini+menggunakan+pendekatan+oversampling+Synthetic+Minority+Oversampling+Technique+(SMOTE)+sebagai+solusi+masalah+data+tidak+seimbang.+Fitur+gambar+diekstraksi+ke+dalam+NumPy+array%2C+dimanipulasi+dan+dianalisis+menggunakan+library+dari+bahasa+pemrograman+Python.+Penelitian+ini+menghasilkan+model+SVM+dengan+akurasi+terendah+yaitu+92.03+dan+akurasi+tertinggi+sebesar+95.90.%0D%0AKata+Kunci+%3A+Dataset+Lampung%2C+Klasifikasi+Gambar%2C+SMOTE+%2C+Support+Vector+Machine%0D%0AImage+classification+is+supervised+learning+by+defining+a+set+of+target+classes%2C+then+training+a+model+using+one+partitioned+dataset+of+training%2C+validation%2C+and+testing+which+contains+labeled+data+of+Lampung+handwritten+images.+The+goal+is+to+develop+a+model+that+could+accurately+classify+these+data+with+the+right+label+using+Support+Vector+Machine+(SVM).+SVM+classification+technique+was+chosen+because+it+could+process+data+in+a+form+of+text+and+images%2C+and+with+the+implementation+of+kernel%2C+data+does+not+have+to+be+linearly+separable+since+input+data+can+be+converted+into+high+dimensional+data.+Information+derived+from+this+endeavor+to+develop+an+optimized+classification+model+can+also+be+used+as+references+for+future+studies+with+a+more+specialized+focus+point%2C+especially+on+similar+subjects+such+as+image+classification.+This+research+uses+traditional+characters+from+a+city+in+Indonesia+called+Lampung.+Unlike+the+conventional+26+Latin+alphabets%2C+Lampung+has+20+characters%2C+12+diacritics%2C+and+unique+punctuations.+Lampung+dataset+consists+of+18+Lampung+characters%2C+in+the+form+of+32140+grayscale+images+without+diacritics+and+punctuations.+These+data+however+are+not+evenly+distributed+among+the+18+characters.+This+research+makes+use+of+oversampling+approach+using+Synthetic+Minority+Oversampling+Technique+(SMOTE)+as+a+solution+for+class+imbalance+found+in+this+dataset.+Lampung+images+features+were+extracted+into+NumPy+array+and+are+manipulated+and+analyzed+using+various+libraries+from+Python+programming+language.+This+research+yields+an+SVM+model+with+the+lowest+accuracy+of+92.03+and+the+highest+accuracy+of+95.90.%0D%0AKeywords+%3A+Image+Classification%2C+Lampung+Dataset%2C+SMOTE%2C+Support+Vector+Machine&rft.publisher=FAKULTAS+MATEMATIKA+DAN+ILMU+PENGETAHUAN+ALAM&rft.date=2022-04-01&rft.type=Skripsi&rft.type=NonPeerReviewed&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F63808%2F1%2FAbstrak.pdf&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F63808%2F2%2FSkripsi%2520Full.pdf&rft.format=text&rft.identifier=http%3A%2F%2Fdigilib.unila.ac.id%2F63808%2F3%2FSkripsi%2520tanpa%2520bab%2520pembahasan.pdf&rft.identifier=++EDO+PRIYONO+MUALIM%2C+1617051113++(2022)+PENDEKATAN+OVERSAMPLING+SMOTE+UNTUK+IMBALANCED+DATASET+AKSARA+LAMPUNG+DAN+KLASIFIKASI+MENGGUNAKAN+SVM.++FAKULTAS+MATEMATIKA+DAN+ILMU+PENGETAHUAN+ALAM%2C+UNIVERSITAS+LAMPUNG.+++++&rft.relation=http%3A%2F%2Fdigilib.unila.ac.id%2F63808%2F