KLASIFIKASI PEKERJAAN SUMUR MINYAK DI PT PERTAMINA HULU ROKAN MENGGUNAKAN METODE RANDOM FOREST
No Thumbnail Available
Date
2023-08
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Elfitra
Abstract
The oil well anomaly detection system has been running for 7 years, generating a large
and complex dataset consisting of indications of oil well issues along with
recommendations for oil well operations by petroleum engineers, commonly referred to
as big data. However, the system has limitations as it can only display indications of well
issues, requiring manual review by petroleum engineers to determine suitable well
operation recommendations based on the data. Additionally, the number of petroleum
engineers at PT Pertamina Hulu Rokan is limited, making it challenging to review all
well issue indications, making it difficult to review all well indications. The objective of
this research is to classify oil well operation data using the Random Forest method. The
random forest model works by constructing multiple decision trees, combining them, and
producing several classes, which are then aggregated to determine the final class.
Performance of the model in classifying oil well operation data using the Random Forest
method with 5 target data labels resulted in an accuracy of 0.66%. The model was
processed by filling missing values with the median, not removing outliers, applying
Oversampling techniques, normalizing the data using MinMax Scaler, and dividing the
data using K-Fold Cross Validation with k=10.
Description
Keywords
Random Forest, Machine Learning, Classification, Oil Well
Citation
Perpustakaan