Putri, ZahraSwastika (2017) Deteksi Autisme Pada Anak Menggunakan Metode Modified K-Nearest Neighbor (MKNN). Sarjana thesis, Universitas Brawijaya.
Abstract
Autisme merupakan gangguan tumbuh kembang anak terkait gangguan komunikasi, kognisi, aktivitas imajinasi dan interaksi sosial. Gangguan tersebut biasanya tampak pada anak sebelum usia 3 tahun. Namun banyak orang tua yang tidak menyadarinya hingga usia 4-7 tahun. Keterlambatan deteksi, kemiripan dengan gangguan tumbuh kembang yang lain dan kurangnya pengetahuan terhadap autisme menyebabkan ketidaktepatan dalam penanganan dan peningkatan jumlah penderita autisme. Identifikasi autisme dibedakan ke dalam autisme berat, autisme sedang, autisme ringan, dan tidak terdeteksi autisme. Metode Modified K-Nearest Neighbor (MKNN) merupakan pengembangan metode KNN konvensional. Proses modifikasi pada MKNN dilakukan dengan menambahkan proses validitas data latih dan proses weight voting. Dengan adanya proses validitas dan weight voting dapat menguatkan ketetanggaan yang ada pada data training serta menguatkan hasil kinerja metode tersebut. Berdasarkan hasil pengujian variasi nilai k didapatkan akurasi tertinggi sebesar 83.33% menggunakan dissimilarity measure. Pada pengujian komposisi keseimbangan data latih didapatkan akurasi tertinggi sebesar 90% menggunakan euclidean distance. Pada pengujian jumlah data latih rata-rata akurasi tertinggi sebesar 79.17%. Pada pengujian variasi data latih didapatkan akurasi tertinggi sebesar 83.33% menggunakan dissimilarity measure. Berdasarkan hasil akurasi pengujian tersebut, menunjukkan bahwa deteksi autisme pada anak menggunakan metode MKNN memiliki tingkat akurasi yang cukup baik dan mampu melakukan klasifikasi deteksi gejala autisme berdasarkan masukan gejala yang dirasakan pengguna.
English Abstract
Autism is a childhood and developmental disorder that is characterized by lack of communication, cognition, imagination and social interaction activities. Symptoms of autism disorder usually appeared and recognized in the first three years of life. But many parents didn’t recognize it until the first four or seven years of life. Delay detection of symptoms, similarities with the other childhood and developmental disorder, and lack of knowledge about autism cause imprecision treatment handling, and increased number of sufferers. Identification of autism differentiated into severe autism, moderate autism, mild autism and non-autism. Modified K-Nearest Neighbor (MKNN) method is a method that enhancing performance of conventional K-Nearest Neighbor method. Process modification on MKNN done by adding validity of the train data process and weight voting process. With that process, MKNN method can robust neighbors in the training data and strengthen the performance results of these methods. Based on variant value of k testing obtained 83.33% accuracy in the dissimilarity measure equation. Based on composition of balance training data testing obtained 90% accuracy in the euclidean distance equation. Based on amount of training data testing obtained 79.17% average accuracy. Based on variation of training data testing obtained 83.33% accuracy in the dissimilarity measure. Based on the results of such testing accuracy, pointed out that the detection of children’s autism using MKNN method have a pretty good degree of accuracy and capable to classify and detection the autism symptoms based on perceived symptoms user input.
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTIK/2017/176/051703832 |
| Subjects: | 000 Computer science, information and general works > 005 Computer programming, programs, data |
| Divisions: | Fakultas Ilmu Komputer > Teknik Informatika |
| Depositing User: | Kustati |
| Date Deposited: | 23 May 2017 08:36 |
| Last Modified: | 22 Oct 2021 03:45 |
| URI: | http://repository.ub.ac.id/id/eprint/147398 |
Preview |
Text
PAPER_Zahra_Swastika_Putri_125150201111025.pdf Download (2MB) | Preview |
Preview |
Text
PASCA_SIDANG-SKRIPSI_Zahra_Swastika_Putri_125150201111025.pdf Download (7MB) | Preview |
Actions (login required)
 |
View Item |