MasteMaster thesis: Exploratory Data Analysis for Enhanced Power System Datar Thesis
Om jobbet
Join our Team
About this opportunity:
This thesis focuses on exploring various data analysing techniques to identify hidden relationships between attributes within datasets. This dataset is a snapshot of the telecommunications power system data where it contains different attributes such as system voltage, system current, battery temperature, etc. The thesis aims to assess how these techniques can aid in improving data modelling. Additionally, the study will evaluate the trade-offs between different analysing techniques (for example, factor analysis - including PCA and maximum-likelihood factor analysis - regression analysis, and time-series analysis). The thesis objectives is to investigate and compare effective data analysis techniques for discovering hidden attribute relationships and patterns in the dataset, specifically focusing on how two or more attributes (e.g., x1, x2, x3) from the power system in site solutions influence or affect another attribute (e.g., y).
What you will do:
This study utilizes telecommunications power system data provided in the form of compressed archive files. Each archive contains a comprehensive snapshot of the power system at a specific point in time, including: time-series measurements from energy meters (voltage, current, power consumption), alarm and event logs, hardware inventory records (device models, serial numbers, revisions), and system configuration data. These snapshots represent the operational state of telecommunications power infrastructure and contain heterogeneous data types-ranging from continuous time-series measurements to structured metadata and discrete event records-that are typically analyzed separately using different tools and methods in current practice.
• Evaluate what are the most effective data-analysis techniques for discovering hidden relationships between attributes in terms of the data from the power system?
• How can we identify patterns in data to improve data modeling or machine learning models?
• Assess which identified patterns are most predictive for downstream tasks (e.g., anomaly detection)?
• Estimate how does the predictive patterns improve the operations site in terms of costs?
The skills you bring:
• Conduct a comprehensive literature review of recent data analysis techniques.
• Implement and evaluate multiple analyzing methods on the consumption dataset.
• Document findings and suggest best practices for data analysis in machine learning workflows.
• Identify some of the patterns that could be identified in the dataset.
About this opportunity:
This thesis focuses on exploring various data analysing techniques to identify hidden relationships between attributes within datasets. This dataset is a snapshot of the telecommunications power system data where it contains different attributes such as system voltage, system current, battery temperature, etc. The thesis aims to assess how these techniques can aid in improving data modelling. Additionally, the study will evaluate the trade-offs between different analysing techniques (for example, factor analysis - including PCA and maximum-likelihood factor analysis - regression analysis, and time-series analysis). The thesis objectives is to investigate and compare effective data analysis techniques for discovering hidden attribute relationships and patterns in the dataset, specifically focusing on how two or more attributes (e.g., x1, x2, x3) from the power system in site solutions influence or affect another attribute (e.g., y).
What you will do:
This study utilizes telecommunications power system data provided in the form of compressed archive files. Each archive contains a comprehensive snapshot of the power system at a specific point in time, including: time-series measurements from energy meters (voltage, current, power consumption), alarm and event logs, hardware inventory records (device models, serial numbers, revisions), and system configuration data. These snapshots represent the operational state of telecommunications power infrastructure and contain heterogeneous data types-ranging from continuous time-series measurements to structured metadata and discrete event records-that are typically analyzed separately using different tools and methods in current practice.
• Evaluate what are the most effective data-analysis techniques for discovering hidden relationships between attributes in terms of the data from the power system?
• How can we identify patterns in data to improve data modeling or machine learning models?
• Assess which identified patterns are most predictive for downstream tasks (e.g., anomaly detection)?
• Estimate how does the predictive patterns improve the operations site in terms of costs?
The skills you bring:
• Conduct a comprehensive literature review of recent data analysis techniques.
• Implement and evaluate multiple analyzing methods on the consumption dataset.
• Document findings and suggest best practices for data analysis in machine learning workflows.
• Identify some of the patterns that could be identified in the dataset.
Ericsson AB
FöretagEricsson AB
Visa alla jobb för Ericsson AB