Data-Driven Insights through Industrial Retrofitting: An Anonymized Dataset with Machine Learning Use Cases.

Area of Science:

• Industrial Engineering
• Data Science
• Machine Learning

Background:

• Small and medium-sized enterprises (SMEs) face challenges in data analysis from retrofitted equipment.
• Existing research inadequately covers data-driven insights in SME environments.

Purpose of the Study:

• To provide an anonymized dataset from SMEs for data-driven analysis.
• To demonstrate the application of machine learning (ML) techniques for extracting insights from limited SME data.

Main Methods:

• Collected anonymized power consumption data from two medium-sized companies over 7 months and 1 year.
• Developed and applied several ML models for tasks including forecasting and classification.
• Utilized a non-invasive and scalable data-collection procedure.

Main Results:

• Demonstrated that ML can extract useful information from limited data types in the short term.
• Successfully applied ML models for power consumption forecasting, item classification, next machine state prediction, and production count forecasting.

Conclusions:

• SMEs can effectively leverage ML techniques with limited data resources.
• The study offers practical insights for SMEs to utilize actionable data insights.
• Findings enhance understanding of ML application in practical SME settings with limited datasets.