Application of Improved Feature Pre-processing Method in Prevention and Control of Electricity Charge Risk

With the increase in power data, past data processing technologies cannot meet the needs of rapid processing and intelligent analysis of massive data. In order to solve the problems of current performance bottleneck in the calculation of power big data and recover users' electricity tariff arre...

Full description

Saved in:
Bibliographic Details
Published in:International journal of performability engineering 2019, Vol.15 (9), p.2453
Main Authors: Huaiguang, Wu, Yongsheng, Shi, Shenyi, Qian, Hongwei, Tao, Jiangtao, Ma
Format: Article
Language:English
Subjects:
Big Data
Cloud computing
Computer networks
Data processing
Distributed processing
Electricity
Electricity generation
Electricity pricing
Regression analysis
Regression models
Risk management
Statistical analysis
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:With the increase in power data, past data processing technologies cannot meet the needs of rapid processing and intelligent analysis of massive data. In order to solve the problems of current performance bottleneck in the calculation of power big data and recover users' electricity tariff arrears in a timely manner, we use the technology of combining parallel computing with feature expansion to establish a feature expansion method based on Spark distributed computing (FESDC). According to the data generated by the proposed method, we use a parallel logistic regression model to predict the arrears probability of future electricity users, which can effectively prevent and resolve the risk of electricity tariff recovery (ETR). Compared with the data processing method for single process (DPSP), the proposed method not only increased the accuracy of prediction, but also improved the performance of processing data.
ISSN:0973-1318
DOI:10.23940/ijpe.19.09.p19.24532461