CD Skripsi

Penerapan Algoritma Lightgbm Dengan Optimisasi Hiperparameter Pada Klasifikasi Emisi Gas Rumah Kaca

---

ABSTRACT
The rise in global greenhouse gas emissions is a serious problem as it can lead to global warming that affects many areas of life, such as droughts and rising seawater. Knowing the classification of future greenhouse gas emissions levels can help in policy planning to minimize the impact. In this study, the classification of greenhouse gas emission levels was done using the lightGBM method by optimizing the learning rate, max depth, and n estimators using the grid search method. The analysis was made using two types of data, namely simulation data and emission data. The simulation was used to evaluate the hyperparameter performance of the lightGBM method, and the greenhouse gas emissions data was used to obtain the best combination in classification. Based on the results of the simulation on the lightgbm method that has been done on the learning rate parameters, the max depth and n estimators for each dataset used do not depend on the height or lowness of the parameter value but instead depend on the correct combination of parameter values to get the best model performance results. Simulation results also showed that increasing the parameter value can initially improve the model's performance, but at certain values, it can lead to a decrease in the performance of such models. In general, the accuracy, specificity, and MAE values do not consistently increase or decrease when the third parameter values are raised, but on the learning rate and max depth parameters, there is a decreasing pattern for the sensitivity values. Based on greenhouse gas emission data, it is obtained that the best parameters are learning rate = 0,1, max depth = 7, and n estimators = 500, with accurate values of 99,15%, sensitivities of 99,81%, and specificity of 97,99%, while MAE is 0,0085.

Keywords: Classification, LightGBM, Grid Search, Hyperparameter, Greenhouse Emissions
 

---

Ketersediaan

Informasi Detail

Judul Seri

-

No. Panggil

2003111099

Penerbit

Pekanbaru : Universitas Riau FMIPA Statistika., 2024

Deskripsi Fisik

-

Bahasa

Indonesia

ISBN/ISSN

-

Klasifikasi

2003111099

Tipe Isi

-

Tipe Media

-

Tipe Pembawa

-

Edisi

-

Subjek

STATISTIKA

Info Detail Spesifik

-

Pernyataan Tanggungjawab

Mutia

Versi lain/terkait

Lampiran Berkas

Komentar

---

Anda harus masuk sebelum memberikan komentar