Skripsi

ANALISIS SENTIMEN TRANSFER PEMAIN SEPAK BOLA MENGGUNAKAN DEEP LEARNING ROBERTA

---

Penilaian

---

Football transfer can impact performance of a club, both on or off the pitch. By analyzing sentiment surrounding player signing, clubs can enhance their marketing strategy to achieve higher earnings. Using 1200 data from comment across some of world class player signing announcements on Twitter, this study aims to perform sentiment analysis task using RoBERTa Model on said data. To adapt RoBERTa model for sentiment analysis, the pretrained RoBERTa base model is modified with additional output layer and then the model is fine-tuned. Testing model revealed that the best accuracy is produced by using epoch value 6 and learning rate value 10-5, resulting in 88,00% accuracy, 89,58% precision, 86,00% recall and 87,75% f-measure. However, there is sign of overfitting during fine-tuning model with learning rate value 10-5 that started to appear at epoch 4. Meanwhile, even though model with learning rate value 10-6 did not indicate any overfit sign, this configuration started slower even in third epoch their accuracy is still only at 50,00%. Eventually, model with learning rate value 10-6 caught up and achieved its best accuracy 85,50% with epoch value 8. Although it is the best for learning rate value 10-6, this accuracy score is still 1,50% behind the lowest accuracy for learning rate value 10-5 that achieve 86,00% accuracy with only 3 epochs. Therefore, it can be concluded that using higher learning rate will result in much faster learning process though it can also drive the model to overfit. Additionally, increase epoch also can increase accuracy but the rate of improvement may vary depending on the learning rate.

---

Availability

Inventory Code	Barcode	Call Number	Location	Status
2307004837	T126992	T1269922023	Central Library (Reference)	Available but not for loan - Not for Loan

Detail Information

Series Title

-

Call Number

T1269922023

Publisher

Indralaya : Jurusan Teknik Informatika, Fakultas Ilmu Komputer Universitas Sriwijaya., 2023

Collation

xvi, 79 hlm.'; Ilus.; 29 cm

Language

Indonesia

ISBN/ISSN

-

Classification

003.07

Content Type

-

Media Type

-

Carrier Type

-

Edition

-

Subject(s)

analisis
Jurusan Teknik Informatika

Specific Detail Info

-

Statement of Responsibility

FIRA

Other version/related

File Attachment

Comments

---

You must be logged in to post a comment