data_analysis

Predictive analysis as a decision model.

KPIs are important tools for an organization to measure the level of performance, taking into account its limitations. The disadvantage of KPi is that they mainly say something about what has happened and less about what may happen.

A good example of predictive analysis is the Google search and autocomplete functions. Based on your argument, Google searches for the most relevant information while you type. This functionality is based on extensive analysis of search arguments and patterns. Google wants to deliver exactly what you are looking for.

We are currently at a crossroads where Big data analysis will drastically change the way organizations handle information. Organizations are currently reporting on what has happened and what is currently going on through real-time information. This is going to shift to what’s going to happen. The purpose of predictive analyzes is to ensure better results, better decisions, insight through relevant information.

How this ultimately translates depends on industry or sector and your information supply chain: raw data, aggregated data, contextual intelligence, insight through analysis, targeted decisions.

Predictive Model.

Predictive models identify and analyze underlying relationships in historical and real-time data. Profiling, categorizing data serves as input for parameters that have a predictive value and thus improve the decision-making process. Such models are already in place in financially sensitive transactions such as international bank transactions.

Both linear and non-linear algorithms provide optimization and insight into raw data. Advanced neural systems learn to recognize complex patterns from social media, for example, and to make predictions based on this. Such as, for example, to monitor and predict disease epidemics.

Learning to decide.

The fact that organizations have a lot of information does not mean that they are successful. It is the quality of analysis and converting it into valuable parameters and the subsequent decisions that determine whether an organization makes optimal use of Big Data. The challenge is: Data * Science * Size * Creativity. But perhaps even more important is whether you have the right foundation in your organization to deal with the predictive information.

Examples

On Time Delivery is a typical supply chain metric. This tells us whether an order was too early or too late depending on the specific criteria. What this KPI did not tell us is why the shipment deviated from the standard. Did this have to do with the supplier, purchasing, product availability [category management] or any other reason. Gaining insight into the development, course and reasons of OTD provides important information. For example, to prevent machine downtime. It can also be used for utilization optimizations or to identify structural problems in your supply chain. Such information influences strategic purchasing decisions.

Project phases.

In the development of your organization to deal with predictive information, there are four basic steps that you go through:

  • Education. Ensuring awareness and focus on new developments and the requirements and benefits for the organization
  • Development. Develop strategy and a roadmap based on business needs and future challenges.
  • To connect. Developing big data pilots that demonstrate added value for your organization and serve as a blueprint for further development of criteria and requirements.
  • Performance. The big data analyzes and results are widely used in the organization when making decisions and further improving the predictive analysis techniques.

 

During the project you will have to make decisions about things like;

  • analysis type. Real time, batch processing, combined
  • Process methodology. predictive analysis of social media, analysis of historical data, development and use of algorithms, reports, translation into decisions
  • Frequencies. On request, continuous, real-time feeding, time batches
  • Data types. meta data, master data, historical, transactions
  • Content. Structured, unstructured or a combination thereof. Produced or derived from images, text, videos, documents, sound recordings
  • Sources. Web and social media, machine generated, human production, internal data, transaction data, biometric, via data providers, originally.
  • Consumer data. human, machine, enterprise applications, repositories
  • Hardware. commodity, internal, cloud, state of the art

Future.

Although predictive data analysis is still in its infancy, it is still growing at a tremendous pace. According to research by Gartner, in 2014, 30% of the analytical applications will make use of proactive, predictive and forecasting competences.
 

http://practicalanalytics.wordpress.com/predictive-analytics-101/
http://www.slideshare.net/perficientinc/predictive-analytics-perficient-perspective#