Data warehousing in Manufacturing Industry- "Record-Keeping of the Hidden Element"

Data Warehousing Manu-minManufacturing industry is very vast and is closely related with IT as well as other departments like material management, product generation, sales and marketing, HR, finance, and so on. All these departments are interlinked and the management becomes easy with the help of various IT systems and services.

Like any other industry, manufacturing also is in need of data logging and accounting based on which business decisions are taken. As time evolves, the amount of data also accumulates. Warehousing or historical record keeping of such data is important for the organization to analyze business situations and understand market trends. Such analysis and decision-making can be achieved using various tools available in data warehousing.

What is data warehousing?            

Data Warehousing (DW) is a computer-based system where historical and transactional data are captured and analyzed in a systematic way so that the data can be used in various analytical and decision-making processes. The process of getting raw data and converting the same into information—to have an in-depth knowledge of the business situation, which leads to creative decision-making—resulting in the generation of better profit is called Business Intelligence (BI).

DW provides various key factors across different dimensions to get the view of the business from the management’s perspective. Replacing the conventional and static reports, these systems use OLAP (On-Line Analytical Processing) applications to generate interactive reports based on the parameters of facts and dimensions.

Basically, manufacturing (or any other) industry has two types of data, dynamic operational data as well as historical data. DW is primarily based on historical data having less number of changes. Such data can be used to perform the following activities:

  • Track performance measures
  • Data analysis
  • Summarize as well as analyze details that provide an outline of all the tasks

Concepts and terminologies in data warehousing

·         Data Warehouse: Is a collection of data which is integrated, subject-oriented, time-variant, and non-volatile. This data supports and influences management’s decision-making process.

·         Data Warehousing: The process of designing, building, and maintaining a data warehouse system.

·         Data Marts: Data mart comprises a subject-oriented set of organizational level data that is created to support analytical reporting requirements pertaining to a specific business unit. It is almost the same as data warehouse, but have more limitations in terms of audience and/or content.

·         Data Cube: A data cube allows us to represent data in various scopes and dimensions. Data cube is described by the parameters of dimensions and facts. The dimensions can be termed as the objects with respect to which an enterprise preserves the records.

·         ETL Framework: This is used to Extract, Transform, and Load data to the data warehouse from multiple heterogeneous source systems.

·         Business Intelligence: Business Intelligence is a logical grouping or arrangement of applications, warehouses, and repositories that are specifically designed for delivering information to the business communities.

·         Dimension: A dimension includes the same category of information. For example: “Time” is a dimension which includes year, month, day, and week.

·         OLAP: On-Line Analytical Processing is based on historical data. OLAP has to be designed in such a way that it provides the end users a quick way to slice and dice the data.

·         Full-Load vs. Incremental Load: Are used to indicate the two different methods of refreshing/loading data to the data warehouse from source systems. In full-load technique, the entire data is flushed out and loaded afresh, whereas in incremental load, only updates are loaded to the system.

Data warehousing Implementation

Implementing a data warehouse system is a strategic decision involving dedicated physical systems and human resources. Typically, a data warehouse is a database server (like Microsoft SQL server) with various data feeding and extracting jobs created inside. DW implementation and operation has three basic steps:

  1. Collect Data

The data collection in DW is commonly named ETL (Extract-Transform-Load) and the following are the various functions in data collection:

  • Data Extraction – Refers to the gathering of data from various source systems. The sources can be multiple homogeneous or heterogeneous systems. For example, sources may include Excel sheets, SQL servers, Oracle database, text files or other sources.
  • Data Cleaning – Is the process of cleaning up the data by finding errors and making corrections. Only the required fields are retained in the data warehouse in the desired form.
  • Data Transformation – Is a process that converts the form of the data from a legacy format to a warehouse format. Irrespective of the heterogeneous nature of the source systems, data is transformed into a uniform format that suits the data warehouse system.
  • Data Loading – Involves merging, sorting, splitting, consolidating, integrity checking, summarizing, logging, and building indices and partitions.
  • Refreshing – Refreshes and updates the data from data sources to warehouse.
  1. Analyze Data

Data analysis can be done either in relational or in multi-dimensional format. Multi-dimensional data analysis structures are based on OLAP concept, where historical data is used instead of the current transactional data. Cube is an example of such a multi-dimensional structure – based on facts and dimensions which include pre-calculating and storing complex aggregations, and building mining models to perform data analysis that helps in identifying information like trends, patterns, relationships, and more.

  1. View/Present Data

Reporting is vital to DW and BI where raw facts are presented as meaningful information in a format liked by the target audience. Such visual analysis of operations, business, and financial data leads to strategic planning and decision-making. Reports can be generated for both active transactional operations as well as passive historical facts.

Various reporting tools are available in the market that provide rich, interactive display with global and mobile access and fast response. Both automatic and manual data refresh facilities are also available.

Tools Available today

There are various tools available in the market to implement data warehousing and business intelligence. Few of them are listed below:

  • Microsoft Business Intelligence

This package is called MSBI and is divided into three categories as given below:

  • SQL Server Integration Services (SSIS) : Extract-Transform-Load operations
  • SQL Server Analysis Services (SSAS): Designing Cubes
  • SQL Server Reporting Services (SSRS) : Operational Reporting
  • SAP Business Objects

This business intelligence suite from SAP provides various components like:

  • Crystal Reports : Operational Reporting
  • Web Intelligence (WebI): Ad-hoc Query and Analysis
  • Explorer : Data Exploration
  • Xcelsius : Dashboards
  • Universes : Database Middleware
  • Query as a Web Service (QaaWS), BI Web Services : Dashboard Integration
  • IBM Cognos

A tool from IBM comprising the features like querying & reporting, analysis, score-carding, dash-boarding, mobile analytics, statistical visualization, real time monitoring, planning, budgets, and so on.

The MicroStrategy platform provides powerful SQL generation, a variety of interfaces for generating and viewing reports, and the ability to customize the functionality and interfaces.

  • QlikView

QlikView offers In-Memory BI platform which is totally different from other conventional query-based BI tools. Here, when the user selects a data point, no queries are fired. Instead of that, all the fields are filtered and reorganized instantly according to the user’s selection.

  • Tableau 

Tableau provides a scalable solution for the creation and delivery of web, mobile, and desktop analytics.

About the author

BincyAfter having vast work experience with DB/BI/Data warehousing side of Automotive and Healthcare domains, Bincy’s new world of exposure is Manufacturing industry – where things behind every product evolves. She is passionate about next gen data mining tools, and is equally crazy about books and photography.

Author : Bincy George Date : 02 Oct 2015