基于特征工程的无线业务感知与KPI指标关联建模研究及应用

Research and Application of Wireless Network KQI and KPI Correlation Modeling Based on Feature Engineering

摘要

针对传统的4G/5G业务感知质差小区优化过程中存在的指标关联断层、决策依赖经验及分析效率低下的问题，引进机器学习技术中的特征重要性选择技术，通过特征重要性评估方法，对KPI特征进行排序和筛选，以确定对业务感知质差影响最为显著的KPI指标。最后，根据特征重要性排名和量化影响程度，为针对性的优化方案的制定提供数据支撑。

Abstract

To address the issues of indicator correlation gaps, reliance on experiential decision-making, and low analytical efficiency in the optimisation of traditional 4G/5G perception-poor cells, the feature importance selection technique from machine learning is introduced. Through feature importance assessment, KPI features are ranked and screened to identify the KPI indicators that have the most significant impact on perception quality. Finally, based on the feature importance ranking and the quantified impact, data support is provided for the formulation of targeted optimisation schemes.

关键词

感知关联 / 特征工程 / 数字化

Key words

perceptual association / feature engineering / digitalisation

参考文献

[1] 吴冬, 阎卫东, 王井利. 基于特征重要性加权的随机森林点云分类研究[J]. 电子测量技术, 2023, 46(20):120-127.DOI:10.19651/j.cnki.emt.2312627.

[2] 罗猛, 李毅, 陈茂林. KQI感知质差小区分析研究[J]. 电信技术, 2018,(03):40-42.

[3] 任小强, 罗志文, 陈苗苗, 等. 基于机器学习算法的5G重点业务KQI和KPI关系研究[J]. 邮电设计技术, 2022,(10):28-31.