22. May 2015

Big Data

What role does „Big Data“ play?

“Big Data” is not simply a buzzword used by high technology companies, but is gaining in recognition in many industries, ranging from retailers to healthcare and manufacturers. The increasing digitalization of processes and the accompanying collection of huge sets of data cause this evolution. By now, any successful company is tracking KPI’s on a regular basis in order to comprehend corporate development. However, the massive amount of data is also causing problems when it comes to processing the data.

In big data applications, seemingly unrelated data pools are fused. Hence, these pools can include internal data (“structured internal data”) as well as external- (“structured external data”) and text-data (“unstructured external data”). Yet, conventional processing methods exhibit limited capabilities on handling these data pools.

Consequently, “Big Data” does not only involve large sets of data but also latest analysis techniques for information processing.

Which analysis methods are employed?

The analysis methods employed by SimCog belong to the area of “predictive analytics”. “Predictive analytics” include statistical methods for pattern recognition, the building of models, and “machine learning”

Building models
Data analyses lead to the identification of key process drivers e.g. the production of consumer products or the daily customer behaviour. The variables identified in this process form a model, which make up the foundation for predictive analytics.

Machine-Learning
Models and algorithms developed by SimCog are always future oriented. That means that our algorithms are set up in an intelligent way in order for them to realign themselves as soon as new data is available. It goes without saying that these processes happen automatically so that no intervention from outside is necessary.

How do you ensure the accuracy of your predictive analytics?

Generally, the data models are created with training and test data sets. That means that the original dataset is split into a training and a test data set.

The model is generated from the training data set and validated against the test data set. If the data is predictable by means of the model, the accuracy of the model is confirmed. Typically, by utilizing these methods, most predictions can be improved by about 40% or more. Simultaneously, the standard deviation is kept to a minimum.

Are there any data security concerns?

SimCog respects your privacy and solely works with anonymized data. The data security laws as well as respect for your privacy is self-evident for SimCog. All data and applications are hosted in the company-owned datacentre in Germany and may not be transferred to third parties. Furthermore, all customer data are saved and protected in an IT-infrastructure that is optimized for machine learning and predictive analytics and secured from third parties. The e-mail correspondence can be secured, of course. We support the usage of cryptographic methods for the confidential transfer of electronic data (e.g. GPG, PGP). In the case of highly sensible data, we can perform an in-house installation of our software into your network in order for the data to never leave your company. For more information click data privacy statement.

SimCog Technologies GmbH
Elbchaussee 1
22765 Hamburg
Germany
info[at]simcog.de
+49 40 4665 658 0
© 2018 SimCog Technologies GmbH