Yuliani, Dyah Dwi (2019) Implementasi Machine Learning Untuk Melakukan Prediksi Hasil Panen Tebu (Studi Kasus di KUD Subur, Malang). Sarjana thesis, Universitas Brawijaya.
Abstract
Tebu (Saccharum officianarum) termasuk ke dalam tanaman jenis rumput-rumputan yang air dari batangnya dimanfaatkan untuk bahan baku gula dan vetsin. Tanaman tebu hanya dapat tumbuh di daerah dengan iklim tropis dengan ketinggian 0-600 mdpl. Produksi tebu di Indonesia pada tahun 2015 tercatat sebesar 2.497.997 ton (Dirjen Perkebunan, 2016). Sentral produksi tebu di Indonesia tahun 2012-2017 terletak di Provinsi Jawa Timur dengan rata-rata produksi 1.186.515 ton atau mencapai 48,26% produksi tebu nasional (Indarti, 2017). KUD Subur mempunyai program unggulan yaitu program Tebu Rakyat Kemitraan (TRK). Dalam pelaksanaannya, program TRK menghasilkan beberapa data yang dirangkum dalam laporan rekapitulasi produksi tebu rakyat kemitraan dalam satu periode musim giling. Data hasil program tersebut apabila digunakan secara tepat dapat digunakan untuk melakukan prediksi hasil panen dan rendemen tebu yang akan didapatkan oleh koperasi dan petani pada tahun-tahun berikutnya, Untuk melakukan pemanfaatan data hasil program Tebu Rakyat Kemitraan, diperlukan implementasi sistem informasi yang dapat menyelesaikan permasalahan-permasalahan tersebut sekaligus memberikan beberapa keuntungan salah satunya adalah meminimasi kesalahan yang sering terjadi dalam proses penginputan data secara manual. Proses perancangan sistem informasi tersebut dapat dilakukan dengan mengimplementasikan Machine Learning. Berdasarkan hasil implementasi machine learning dan pebandingan beberapa algoritma, menunjukkan bahwaalgoritma paling baik yang menghasilkan prediksi paling baik adalah agoritma gradient boosting dengan akurasi moel 0.78%. sistem prediksi hasil panen kemudian diuji menggunakan pengujian black box dan user acceptance testing. Didapatkan hasil pengujian black box 95% berhasil yang menyatakan bahwa sistem dapat memenuhi kebutuhan pengguna. Pada pengujian user acceptance testing, didapatkan hasil keseluruhan sebesar 87.5% yang menggambarkan pengguna daapt menerima dengan baik sistem dalam tiga kriteria yaitu performance, usability, dan accuracy.
English Abstract
Sugar cane ( Saccharum officianarum ) is included in the grass-type plants whose water from the stems is used as raw material for sugar and vetsin. Sugarcane can only grow in a tropical climatem areas with an altitude of 0-600 meters above sea level. Sugar cane production in Indonesia in 2015 was recorded at 2,497,997 tons (Director General of Plantations, 2016). Central sugarcane production in Indonesia at 2012-2017 is located in East Java Province with an average production of 1,186,515 tons or 48.26% of national sugarcane production (Indarti, 2017). KUD Subur has a superior program called the People's Sugar Cane Partnership Program (TRK). In practice, the program TRK generate some data summarized in summary report sugarcane production partnership over a period of grinding season. If the program results data is used appropriately, it can be used to predict yields and yields of sugarcane which will be obtained by cooperatives and farmers in the following years. For the utilization of data from the Sugar Cane Partnership program, it is necessary to implement an information system that can solve these problems while providing several advantages, one of which is to minimize errors that often occur in the process of manually inputting data. The information system design process can be carried out by implementing Machine Learning. Based on the results of the implementation of machine learning and comparison of several algorithms, it shows that the best algorithm that produces the best predictions is the gradientboosting algorithm with an accuracy of 0.78% moel. Then, the yield prediction system is tested using black box testing and user acceptance testing. The black box testing showed 95% succeeded in stating that the system can meet the needs of users. In the user acceptance testing, the result overall amounted to 87.5% which illustrates well the user can accept the system in three criteria: performance, usability, and accuracy
Other obstract
-
| Item Type: | Thesis (Sarjana) |
|---|---|
| Identification Number: | SKR/FTP/2019/319/052002280 |
| Uncontrolled Keywords: | Hasil Panen, Machine Learning, Tebu, Yields result, machine learning, sugarcane |
| Subjects: | 600 Technology (Applied sciences) > 633 Field and plantation crops > 633.6 Sugar, syrup, starch crop > 633.61 Sugarcane |
| Divisions: | Fakultas Teknologi Pertanian > Teknologi Industri Pertanian |
| Depositing User: | Nur Cholis |
| Date Deposited: | 18 Aug 2020 03:19 |
| Last Modified: | 21 Oct 2021 07:10 |
| URI: | http://repository.ub.ac.id/id/eprint/181481 |
Preview |
Text
Dyah Dwi Yuliani. (2).pdf Download (5MB) | Preview |
Actions (login required)
 |
View Item |