Strategi Peningkatan Pencapaian SLA Berbasis Eksplorasi dan Peramalan Komplain Pelanggan Infrastruktur Telekomunikasi di PT XYZ

Ahmadi, Mohammad Syauqi (2026) Strategi Peningkatan Pencapaian SLA Berbasis Eksplorasi dan Peramalan Komplain Pelanggan Infrastruktur Telekomunikasi di PT XYZ. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6032221174-Master_Thesis.pdf Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Penanganan komplain pelanggan merupakan indikator krusial dalam penilaian kinerja PT XYZ sebagai penyedia infrastruktur telekomunikasi berbasis Business-to-Business (B2B). Saat ini, terdapat tantangan operasional di mana tingkat kegagalan pemenuhan target waktu penyelesaian (Service Level Agreement / SLA) komplain mencapai 9,5%, yang melampaui batas toleransi dari target pencapaian perusahaan sebesar 99%. Keterlambatan ini berisiko menimbulkan penalti operasional finansial serta menurunkan kepercayaan pelanggan. Penelitian ini bertujuan untuk mengidentifikasi penyebab utama komplain melalui Exploratory Data Analysis (EDA) menggunakan Analisis Pareto dan analisis temporal, serta menerapkan pendekatan pemodelan time series komparatif antara Autoregressive Integrated Moving Average (ARIMA) dan Seasonal Autoregressive Integrated Moving Average (SARIMA) untuk memproyeksikan volume komplain harian. Berdasarkan Analisis Pareto terhadap 12.513 tiket komplain periode Januari hingga Oktober 2023, gangguan Catu Daya PLN menjadi penyebab paling dominan (vital few) dengan kontribusi mencapai 82,30% (10.298 tiket). Evaluasi analisis temporal mengonfirmasi adanya pola musiman mingguan (s=7) dengan puncak komplain terjadi pada hari Selasa (51,84 tiket/hari) dan terendah pada hari Minggu (28,25 tiket/hari). Hasil pengujian out-of-sample pada data testing (1–31 Oktober 2023) menunjukkan bahwa model musiman SARIMA(1,0,1)(1,0,1)7 menghasilkan kinerja peramalan terbaik dalam menangkap fluktuasi mingguan secara dinamis dengan nilai akurasi relatif MAPE terkecil sebesar 27,72% dan MAE sebesar 13,35 tiket/hari, mengungguli model baseline ARIMA(1,0,1) yang menghasilkan garis proyeksi datar. Proyeksi dari model SARIMA ini menjadi dasar perumusan rekomendasi strategi operasional melalui tiga pilar utama: penguatan keandalan catu daya (Lithium-ion battery, Mobile Genset, ATS, dan NMS), alokasi dinamis personel teknis siaga (dynamic shift scheduling), serta integrasi Early Warning System (EWS) pada dashboard NOC guna meningkatkan ketepatan penanganan SLA dan meminimalkan denda operasional pada PT XYZ.
=====================================================================================================================================
Customer complaint handling serves as a crucial performance indicator in evaluating PT XYZ as a Business-to-Business (B2B) telecommunications infrastructure provider. Currently, operational challenges have been identified where the failure rate to meet the complaint resolution time frame (Service Level Agreement / SLA) reached 9.5%, exceeding the allowable tolerance threshold set by the company's target achievement rate of 99%. These delays pose a direct risk of financial penalties and degrade customer trust. This study aims to identify the root causes of complaints through Exploratory Data Analysis (EDA) using Pareto Analysis and temporal analysis, as well as applying a comparative time-series forecasting approach between Autoregressive Integrated Moving Average (ARIMA) and Seasonal Autoregressive Integrated Moving Average (SARIMA) to project daily complaint volumes. Based on the Pareto Analysis of 12,513 complaint tickets from January to October 2023, Power Grid (PLN) outages emerged as the single most dominant cause (vital few), contributing 82.30% (10,298 tickets) of total cases. Temporal evaluation confirmed a distinct weekly seasonal pattern (s=7), with complaint volumes peaking on Tuesdays (51.84 tickets/day) and reaching their lowest point on Sundays (28.25 tickets/day). Out-of-sample testing evaluation on the test dataset (October 1–31, 2023) demonstrated that the seasonal SARIMA(1,0,1)(1,0,1)7 model achieved the best predictive performance in dynamically capturing weekly fluctuations, achieving the lowest relative error with a MAPE of 27.72% and a MAE of 13.35 tickets/day, outperforming the baseline ARIMA(1,0,1) model which produced a flat forecast trajectory. The forward forecast generated by the SARIMA model provides the baseline for strategic operational recommendations across three key pillars: power reliability mitigation (Lithium-ion batteries, Mobile Gensets, ATS, and NMS integration), dynamic shift scheduling for field engineers, and the integration of an Early Warning System (EWS) into the NOC dashboard to enhance SLA resolution ratios and minimize operational penalty risks for PT XYZ.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Infrastruktur Telekomunikasi, Komplain Pelanggan, Analisis Pareto, Time Series Forecasting, ARIMA, SARIMA, Service Level Agreement (SLA), Telecommunications Infrastructure, Customer Complaints, Pareto Analysis, Time Series Forecasting, ARIMA, SARIMA, Service Level Agreement (SLA)
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Interdisciplinary School of Management and Technology (SIMT) > 61101-Master of Technology Management (MMT)
Depositing User:	Mohammad Syauqi Ahmadi
Date Deposited:	05 Aug 2026 02:18
Last Modified:	05 Aug 2026 03:02
URI:	http://repository.its.ac.id/id/eprint/143831

Actions (login required)

View Item