Slide 1
Slide 1 text
Mengoptimalkan
Pengalaman Pengguna
dengan Sistem
Rekomendasi
Machine Learning
Mochamad Rafy Ardhanie
Curriculum Developer
Slide 2
Slide 2 text
Key Point
Machine Learning
● Pengenalan sistem rekomendasi
● Ragam sistem rekomendasi
● Content based filtering
● Collaborative filtering
● Hybrid Recommender
● Be a friend with math
● Code for lyfe
Slide 3
Slide 3 text
Machine Learning
Slide 4
Slide 4 text
Machine Learning
Slide 5
Slide 5 text
Machine Learning
Slide 6
Slide 6 text
Machine Learning
Slide 7
Slide 7 text
Pengenalan Sistem
Rekomendasi
Machine Learning
Slide 8
Slide 8 text
Machine Learning
Sistem Rekomendasi
A system that predicts ratings or
preferences a user might give to a
product. Often these are sorted and
presented as “top-N''
recommendations. Also known as
recommender engines,
recommendation systems, and
recommendation platforms
Slide 9
Slide 9 text
Machine Learning
Mengapa Sistem Rekomendasi itu Penting?
Slide 10
Slide 10 text
Machine Learning
Perusahaan Pengguna
Slide 11
Slide 11 text
Machine Learning
Slide 12
Slide 12 text
Machine Learning
https://atrium.ai/resources/what-are-recommendation-systems-and-how-are-they-transforming-our-markets/
Slide 13
Slide 13 text
Machine Learning
Slide 14
Slide 14 text
Machine Learning
Meningkatkan
Penjualan
Mengapa Sistem Rekomendasi itu Penting?
Penjualan
Beragam
Kepuasan
Pelanggan
Personalisasi
Slide 15
Slide 15 text
Ragam Sistem
Rekomendasi
Machine Learning
Slide 16
Slide 16 text
Machine Learning
Slide 17
Slide 17 text
Machine Learning
Non-Personalized Recommender System
Slide 18
Slide 18 text
Machine Learning
Best-Seller
People who X also Y
Most Popular
Best-Liked
Trending Hot
Non-Personalized
Recommender
System
Slide 19
Slide 19 text
Machine Learning
Personalized Recommender System
Personalized
Collaborative Filtering
Content-Based
Filtering
User Based Item Based
Hybrid Method
Matrix Factorization Deep Learning
Memory Based Model Based
Slide 20
Slide 20 text
Content-Based Filtering
Machine Learning
Slide 21
Slide 21 text
Machine Learning
Content-Based Filtering
Slide 22
Slide 22 text
Machine Learning
Content-Based Filtering
Slide 23
Slide 23 text
Machine Learning
Content-Based Filtering
A.B = (1*1)+(1*1)+(1*1)+(0*0)+(0*0)
||A|| = √12+12+12 =1
||B|| = √12+12+12 =1
Cosine Similarity = 1
Slide 24
Slide 24 text
Machine Learning
Content-Based Filtering
Genre: Action = [1, 0], Adventure = [0, 1]
Platform: PC = [1, 0], Console = [0, 1]
Rating: nilai rating itu sendiri
Game A: [1, 0, 1, 0, 4.5]
Game B: [1, 0, 0, 1, 4.0]
Game C: [0, 1, 1, 0, 3.5]
Slide 25
Slide 25 text
Machine Learning
Content-Based Filtering
Slide 26
Slide 26 text
Machine Learning
Content-Based Filtering
Slide 27
Slide 27 text
Collaborative Filtering
Machine Learning
Slide 28
Slide 28 text
Machine Learning
Collaborative Filtering
Slide 29
Slide 29 text
User-Based
Slide 30
Slide 30 text
Item-Based
Slide 31
Slide 31 text
Model-Based
Cluster Based
Matrix
Factorization
Deep Learning
Slide 32
Slide 32 text
Hybrid Recommender
System
Machine Learning
Slide 33
Slide 33 text
No content
Slide 34
Slide 34 text
Machine Learning
Slide 35
Slide 35 text
Hybrid Recommender System
Mixed Weight Switching
Slide 36
Slide 36 text
Be a Friend with Math
Machine Learning
Slide 37
Slide 37 text
Similarity Measure
Bagaimana Anda mendefinisikan
kesamaan (kemiripan) ini?
Kesamaan paling umum yang bisa
diukur dalam sistem rekomendasi
antara lain, preferensi dan selera.
Selain itu kesamaan juga dapat
ditemukan melalui data dan
informasi lain seperti demografi
pengguna dan status sosial.
Slide 38
Slide 38 text
Jaccard Distance
Teknik ini menghitung seberapa dekat jarak antara dua set sampel. Dalam hal ini, set
sampel bisa berupa sekumpulan restoran yang disukai pengunjung. Koefisien ini
mengukur kesamaan dengan menghitung irisan antara sampel dibagi dengan
gabungan sampel.
Slide 39
Slide 39 text
Cosine Similarity
Cosine similarity mengukur kesamaan antara dua vektor dan menentukan apakah kedua vektor tersebut
menunjuk ke arah yang sama. Ia menghitung sudut cosinus antara dua vektor. Semakin kecil sudut cosinus,
semakin besar nilai cosine similarity.
Slide 40
Slide 40 text
Euclidean Distance
Metode Euclidean distance dihitung berdasarkan jarak antara dua titik (item). Item serupa letaknya akan
berdekatan satu sama lain jika diplot dalam ruang berdimensi-n.
Slide 41
Slide 41 text
Code for Lyfe
Slide 42
Slide 42 text
Thank You
Rafyardhani [email protected]
Rafyardhani
RafyArdhanie
Machine Learning
Slide 43
Slide 43 text
Feedback!
Hadiah:
● 1 Token Langganan Academy 30 Hari)
*untuk pengisi feedback terpilih!
dicoding.id/devcoachfeedback