Tips and tricks for a successful data analytics project

Tips and tricks for a successful data analytics project

machine data, service products

Tips and tricks for a successful data analytics project

The digitalization of mechanical production and manufacturing processes is becoming increasingly important. In the meantime, every 4th machine is smart and works in a network.  With the help of sensors and Big Data analyses, the factory is becoming smart production. Machines, products, customers and suppliers are networked with each other as well as externally. Costs are reduced and processes are optimized through digitalization. With the help of Big Data Analytics, companies should be able to make better business decisions.

With the enormous amount of data that a machine generates, the relevant information must be obtained from the large, unstructured set. Big Data Analytics is operated using software tools. When implemented correctly, costs can be saved and your own production can be better understood and optimized. A prerequisite for this are IT infrastructures that enable the high-performance storage and processing of large amounts of data.

Conditions & challenges

With the help of a Big Data Analytics project, existing business processes can be optimized, and the efficiency of data processing increased. Even previously unused data can be used to support processes. However, whether a Big Data Analytics project will be of great success and benefit is difficult to assess in advance. The reason for this are the few empirical values regarding the effort and benefits.

In order to introduce digitization into the company, new structures of cooperation must be created. New ways of thinking are also often necessary. The requirements are also defined by customer needs. Since digitization is associated with greater complexity, close cooperation with specialists from a wide range of specialist areas is crucial. Strategies must be continuously rethought and adapted to customer needs.

Use Cases and Smart Services

Let us assume that we only want to extract the salt from the saltwater. It works similarly with Big Data analysis. Here information is extracted from a large amount of data and used to generate new products and digital business models. This can be, for example, the automatic detection of machine errors, requesting the service technician via digital systems or AI-supported condition monitoring to monitor the chronological development of states of critical machine components.

Use cases are required for a successful implementation of the Big Data Analytics project. Use cases are use cases. The creation of use cases is possible using Smart Services. Smart Services are analysis algorithms that extract useful information from data. There are three different categories of Smart Services: technical, economic and sales Smart Services. The usage behaviour of the customer is the focus of sales-related Smart Services. An example of this would be the calculation of recommendations for the purchase of machine-specific consumables. In the case of economical Smart Services, the focus is on software-driven optimization of production processes, such as reducing waste through optimized machine configuration. Technical Smart Services are concerned with monitoring the functionality of machines, such as detecting errors in production processes or predicting the wear level of machine components. However, several Smart Services are usually involved in the implementation.

Creation of Use Cases

Effective use cases create real added value. Therefore, the focus should initially be on the following question: "Which data-driven products and business models do the customer or the own company need and which models could be relevant in the future? For the development of ideas and solution approaches it makes sense to bring together employees from different departments. Knowledge of internal processes and their weaknesses, background knowledge about customers and their needs, technical expertise, strategic knowledge and creativity are helpful for a good use case. Ideally, several Use Cases are developed and sorted according to their priority.

Traps and how to avoid them

Many things can go wrong with the implementation. Possible risk factors in the development of use cases could be the lack of support from management, lack of understanding and lack of acceptance by employees. In addition, use cases must be realizable. Too little information in the data is the most common obstacle.

For this reason, a research phase in advance is indispensable. The aim of this phase should be to ensure that all the necessary information is included in the data for implementing the use case. If the information content of the data is not enough, the data situation must be improved by appropriate measures. This can be done, for example, by using additional data sources or by improving sensor technology. If this was also not promising, the project is terminated and the next use case is investigated. However, it should be noted that as the ranking of the use case declines, the relevance for the customer also decreases.

Another important point is that the use cases must be geared to the customer's requirements (added value and integrability into existing workflows). Added value is created when the customer can integrate the use case into his production processes. If customer data is required for the development of a use case, sufficient customers must be prepared to provide this necessary data.

Data quality

The most important element in a Big Data Analytics project is the data. The quality of the data is more important for the success of the use case than the pure quantity of data. But how is the quality of the data recognized? Data quality only arises in connection with the respective use case. It is necessary to answer the question: "Is there enough information in the form of trends, patterns and correlations in the data?

The amount of information is determined by two factors: the choice of data sets or measured variables and by the measurement process itself. The first thing to do is to identify the data correctly. Measurement technology and data science are used for this. Measurement technology enables data to be recorded using suitable hardware and made available to other systems. Data science then draws conclusions from the data and generates knowledge, for example by means of certain analysis algorithms.

IT infrastructure

Another important point in the implementation of Big Data Analytics projects is the development of IT systems such as database systems or information systems. These enable the enormous amount of data to be stored and processed. In addition, it must be possible to integrate the data precisely into existing IT systems. Only in this way can the added value generated be poured into existing workflows, products or business models. Before implementation can begin, a suitable architecture model that meets the requirements must be found.

In general, IT systems have to meet a wide range of requirements, such as storing and processing large, unstructured volumes of data or analyzing small amounts of data in real time. In addition, a constant adaptation of the system to scalability is necessary to meet future requirements. It is important to do this right from the start, so as not to rely on a system that cannot be further implemented under heavy load.

IT Security

If a Big Data platform is used, various risks can arise. Various factors play a role here: for example, attacks from outside, incorrect use of the systems or human error.

Nevertheless, data should be stored as securely as possible to protect against data misuse, theft and the associated loss of trust.

There is no generally valid approach to this problem. But there are a few things that should always be considered:

  • Communication of all systems only via encrypted channels
  • Distribute data to several instances. This prevents failures. If, for example, a subsystem fails, this affects performance, but does not lead to a total failure.
  • In addition, data should be protected by a suitable backup strategy with regular backups if it is written incorrectly, deleted, or if a software error occurs.

Successful implementation of the Big Data Analytics project

Creativity, collaborative, multidisciplinary and agile work play an important role in the implementation of Big Data Analytics projects. For successful projects, meaningful use cases, analysis algorithms, an IT infrastructure and correct data are the prerequisites. In addition, customers and employees should be involved early enough in the development of products and services, because they decide whether a solution will be successful or not. Especially important is: Get started first!