@mastersthesis{digilib44369,
           month = {January},
           title = {KOMPARASI ALGORITMA KNN DAN ANN PADA PREDIKSI
HASIL STUDI MAHASISWA 
(STUDI KASUS: MAHASISWA UNIVERSITAS AHMAD DAHLAN YOGYAKARTA)},
          school = {UIN SUNAN KALIJAGA YOGYAKARTA},
          author = {NIM.: 18206050005 Sugriyono},
            year = {2021},
            note = {Pembimbing : Maria Ulfah Siregar, S.Kom. MIT., Ph.D.},
        keywords = {Scikit-learn, Confusion Matrix, K-fold Cross Validation, k-Nearest  Neighbor  (kNN) , Matriks Jarak,  K-means},
             url = {https://digilib.uin-suka.ac.id/id/eprint/44369/},
        abstract = {Classification techniques in data mining can be used to make
predictions. ANN and kNN algorithms are classification methods that
are widely implemented in prediction of student study outcomes. Many
implementations  of  this  classification  method  have  low  accuracy
results. One of the causes that influence low accuracy of classification
results is the quality of dataset. This study aims to compare ANN and
kNN  classification  algorithms  to  determine  the  best  classification
method of two algorithms in predicting student study outcomes. Preprocessing
is
carried
out
to
improve
data
quality
by
removing
outliers
from
datasets
that
already
have
label
classes.
The
student
study
result
dataset
 is  6.847  instances,  with  18  attributes  and  three  classes.  Kmeans
algorithm
and
distance
matrix
are
used
for
preprocessing.
Kmeans
to
get
cluster
center
in
each
class,
the
distance
matrix
is
used
to
evaluate
 distance  of  instance  from  cluster  center.  Euclidean  and
Manhattan  distance  matrices  are  used  as  a  comparison  of  the
performance of distance matrix. Based on the research results, ANN
algorithm  has  greater  accuracy  than  kNN  algotithm,  with  K-means
preprocessing method and Euclidean distance matrix. Using 10-fold
cross validation and confusion matrix  test method, ANN algorithm has
98,99\% accuracy, while kNN algorithm has 98,45\% accuracy.}
}