About this lab
The lab focuses on Analytic discovery from diverse sources of manufacturing data leveraging a product called FactoryTalk® Analytics™ Data Explorer. Analytics is the discovery of information from data and frequently leveraging new and intelligent methods to integrate diverse sources of information, which could include process historical data, manufacturing execution system (MES) data, maintenance management system (MMS) data, laboratory information management systems (LIMS) data, enterprise resource planning (ERP) data among others.
Discoveries and analytic applications are based on two different strategies in general, observation and execution of logic versus streaming data (run-time) and interaction with past data (batch). Both add value and provide insights in different ways, but frequently this review of already available past data is a rapid way to identify interactions and 'what happened'. Thus in many activities analytics starts by reviewing batch historical information (where streaming information is now and in the future).
Cross Industry Standard Process for Data Mining (CRISP-DM)
The above image represents the Cross Industry Standard Process for Data Mining, to indicate how analytic investigations start by evaluating past data, identifying and aligning relevant data with business targets, preparing that data for discovery and occasionally result in a mathematical model (machine learning) from data that provides deeper insights and can be run in parallel (streaming) with manufacturing operations (or any system).
Surveys of professional data scientists indicate that most of their time is consumed by the preparation, merging and clean-up of data to make it relevant for discovery and machine learning. Data Explorer is focused on this exercise. Data Explorer supports:
- Merging data from different resources with a flexible, powerful merge function.
- Graphically eliminating bad, irrelevant or undesired data in a very simple toolset.
- Segmenting interesting data between grades, operating conditions/status.
- Generating data enrichment functions as understood or desired (dynamic filtering, physical expressions).
- Graphically identifying interactions, correlations and exploring time-relevant interactivity between data
- Exporting data for use by other tool sets.Data Preparation Survey
Data Explorer is a standalone product, which is designed for engineers to prepare and interact with their own data. This includes exporting data for external machine learning algorithms. In addition it is being built into integrated machine learning toolsets focused on Predictive KPIs, Process Anomaly Detection and Predictive Maintenance also designed for self-service analytics with similar interaction with data toward development of streaming applications to run within FactoryTalk® Analytics™. Teams of engineers can share work where the right expert on their business opportunities can make intelligent decisions on their own data and continue to complete steps appropriate for specific business problems.
Provide Feedback