Interview About Data Analytics and Its Trends

8th November 2019 ǀ 9 minutes

MR. TOMÁŠ MUŽÍK, OUR DELIVERY MANAGER, REVEALS WHY IT IS USEFUL TO PAY ATTENTION TO DATA ANALYTICS, WHERE POTENTIAL PITFALLS CAN BE FOUND AND WHERE ITS DEVELOPMENT MIGHT HEAD IN THE FUTURE.

Interview About Data Analytics and Its Trends

8th November 2019 ǀ 9 minutes

MR. TOMÁŠ MUŽÍK, OUR DELIVERY MANAGER, REVEALS WHY IT IS USEFUL TO PAY ATTENTION TO DATA ANALYTICS, WHERE POTENTIAL PITFALLS CAN BE FOUND AND WHERE ITS DEVELOPMENT MIGHT HEAD IN THE FUTURE.

Since companies in today’s digital era generate vast amounts of data, an opportunity presents itself to extract useful insights from this information. Data Analytics is a discipline that uses various methods to derive valuable insights from data and help companies run more efficiently.

In the following interview, Mr. Tomáš Mužík, our Delivery Manager, reveals why it is useful to pay attention to data analytics, where potential pitfalls can be found and where its development might head in the future.

What type of companies is data analytics suitable for?

Basically, it can be used in any company since there virtually isn’t a single one that doesn’t leave a digital footprint through its actions. If I were to name the industry sectors for which data analytics is important, I would definitely mention banks and insurance companies, or more generally financial service businesses.

It may appeal to manufacturing companies that – in addition to transaction data – also have technical data available to be processed and utilised. For instance, one can look for machines that are expected to break down and repair them in the near future, change spare parts in advance, both leading to cost savings.

All network companies, i.e. telecommunications and energy production companies, are also suitable clients. Public administration, which has huge amounts of the citizens’ personal data and life situations, can also be added to the list.

Tomáš Mužík | Head of Delivery, Ness Digital Engineering

How to implement data analytics effectively?

With regards to data analytics, most companies have already implemented it in some form. I hardly ever encounter the situation of a company starting from scratch in data analytics.

They say: think big, start small. That means you have to think of an overall concept that you want to achieve. Data analytics is not primarily based on the technologies you use, nor how you build a data model or how often you “serve out” data. It instead relies on teams and processes that are built around the data and are able to figure out what is to be searched in the data.

Who should decide on the implementation?

Most of the activities that are carried out in the context of data analytics are focused on two areas. The first one is the regulatory agenda, primarily in the financial and utilities field. The second area is devoted to customer behaviour analysis and is driven by an effort to offer the right products to the customer, to extend the time the customer cooperates with the company (for example, purchases), to attract new customers and to maximise margins.

Management, of course, is the decision maker in the regulatory administration. Regarding the secondary administration, it is Sales or Customer Relationship Management that matters more.

Where can companies encounter troubles when implementing data analytics?

Clearly, data availability and quality are key. There are plenty of frameworks explaining how to work with Master Data Management or Data Governance, how to clean data, how to maintain it at a reasonable quality, etc. But only very few companies have procedures in place to cover all the risks in some way.

All the information you extract is as good as the quality of your input data. However, the cost of data management is relatively high, and hardly any companies count on it when starting a data analytics project.

What are the benefits of using data analytics?

As a general rule, it comes down to better decision making. This is a little clichéd, so let me explain more specifically.

In the area of customer analysis, the benefit is, for instance, a significant increase in conversion in targeted marketing campaigns. If you correctly select the ones to be addressed from your customer list, you will get a better response in terms of conversion that people actually buy your product, compared to the situation of not managing the process and “shooting” blindly at random.

Regarding the regulatory area, the benefit is a reduction in all possible risks. The result is, for example, better auditability.

I have recently heard of an excellent example of optimising power consumption in data centres. Using unsupervised learning (a method where a computer figures out for itself some relationships via a deep neural network without explicit instructions) to check the data centre’s workload, the computer was shutting off or switching on power supplies (computers, disk arrays etc.). Within a short period, it managed to find a stable model that did not affect the quality of service for clients, yet via targeted shutdown of unused sources or devices it reduced energy consumption by 30%, which is a particularly important fact and savings for data centres, as these have really enormous energy consumption. This is an interesting benefit.

How to optimise processes using data analytics?

As I said, all business activities nowadays leave a digital footprint. Take a look at the simple example of invoice delivery. You know that, when you have received an invoice, someone has scanned and forwarded it before. Each step has its timestamp. Based on the transaction logs, you can see where the invoice and other documents are moving, whether the trajectory is optimal or whether it can be improved. This is an area where data analytics is not yet fully utilised. In my opinion, this is a future direction that is worthwhile considering.

What are the trends in Data Analytics?

I’d like to mention attempts to use unstructured data, and there are only a few real business examples.

In addition, I’d point to sensible use of machine learning represented by the example of data centre power consumption control. However, machine learning is not suitable for many tasks because you are missing a description of causality. For a better understanding, let me give you an example of an insurance company that has some fraudulent cases in its stack of insurance claims.

You need to identify them, pass them on to investigators and decide in specific cases whether it’s fraud or not. These people need to know why you think it’s fraud. But the neural network won’t tell you. It just carries out calculations. It suggests the cases, but it doesn’t say it explicitly as it doesn’t know any facts that clearly show fraud. So follow the trends a bit carefully and apply them only where it makes sense.

It is worth mentioning self-service data analytics. Increasingly, businesses are striving to transfer capability and responsibility for data analysis to the end user, rather than leaving it to their IT department or external organisation. This has some benefits, typically the speed of processing, and information interpretation. I would call it the democratisation of Data & Analytics.

Another example is the Internet of Things (IoT). I might have a lot of sensors that travel around the country or the world in various ways, and I would concentrate their data in one place and extract them. A typical concept in this case is cloud processing, i.e. the trend of Data & Analytics in the cloud.

Crowdsourcing is also an interesting topic. I know a company that operates in the field of loans. It has developed credit risk models that tell it to whom it can lend and to whom it shouldn’t. These models have been refined over time. This company wanted to move on, so it used crowdsourcing.

They announced a world-wide competition, with payment of course, to find out who would come up with the model that best describes the situation of “bad payers”. The competition subscribers got access to anonymised data on those clients who paid and those who had a problem with payment. It really worked and they managed to improve the model.

What would you like to add at the end?

Technological tools are great, and there are many things to do in data analysis, but it really depends on the invention of people working with them in the company and their motivation.

I had experience of a project -built on an amazing data warehouse that covered virtually all the company’s data sources. The first two tasks were implemented – controlling reports and data reconciliation between the accounting and operational systems and some marketing stuff, and that was the end of the story.

The people who started the project were full of enthusiasm. Unfortunately when they left the company, the continuity was broken. It’s always a question of people. Technology and processes can only become truly useful when used wisely by the people.

Tomáš Mužík
Head of Delivery, CZ & SK

Since companies in today’s digital era generate vast amounts of data, an opportunity presents itself to extract useful insights from this information. Data Analytics is a discipline that uses various methods to derive valuable insights from data and help companies run more efficiently.

In the following interview, Mr. Tomáš Mužík, our Delivery Manager, reveals why it is useful to pay attention to data analytics, where potential pitfalls can be found and where its development might head in the future.

What type of companies is data analytics suitable for?

Basically, it can be used in any company since there virtually isn’t a single one that doesn’t leave a digital footprint through its actions. If I were to name the industry sectors for which data analytics is important, I would definitely mention banks and insurance companies, or more generally financial service businesses.

It may appeal to manufacturing companies that – in addition to transaction data – also have technical data available to be processed and utilised. For instance, one can look for machines that are expected to break down and repair them in the near future, change spare parts in advance, both leading to cost savings.

All network companies, i.e. telecommunications and energy production companies, are also suitable clients. Public administration, which has huge amounts of the citizens’ personal data and life situations, can also be added to the list.

Tomáš Mužík | Head of Delivery, Ness Digital Engineering

How to implement data analytics effectively?

With regards to data analytics, most companies have already implemented it in some form. I hardly ever encounter the situation of a company starting from scratch in data analytics.

They say: think big, start small. That means you have to think of an overall concept that you want to achieve. Data analytics is not primarily based on the technologies you use, nor how you build a data model or how often you “serve out” data. It instead relies on teams and processes that are built around the data and are able to figure out what is to be searched in the data.

Who should decide on the implementation?

Most of the activities that are carried out in the context of data analytics are focused on two areas. The first one is the regulatory agenda, primarily in the financial and utilities field. The second area is devoted to customer behaviour analysis and is driven by an effort to offer the right products to the customer, to extend the time the customer cooperates with the company (for example, purchases), to attract new customers and to maximise margins.

Management, of course, is the decision maker in the regulatory administration. Regarding the secondary administration, it is Sales or Customer Relationship Management that matters more.

Where can companies encounter troubles when implementing data analytics?

Clearly, data availability and quality are key. There are plenty of frameworks explaining how to work with Master Data Management or Data Governance, how to clean data, how to maintain it at a reasonable quality, etc. But only very few companies have procedures in place to cover all the risks in some way.

All the information you extract is as good as the quality of your input data. However, the cost of data management is relatively high, and hardly any companies count on it when starting a data analytics project.

What are the benefits of using data analytics?

As a general rule, it comes down to better decision making. This is a little clichéd, so let me explain more specifically.

In the area of customer analysis, the benefit is, for instance, a significant increase in conversion in targeted marketing campaigns. If you correctly select the ones to be addressed from your customer list, you will get a better response in terms of conversion that people actually buy your product, compared to the situation of not managing the process and “shooting” blindly at random.

Regarding the regulatory area, the benefit is a reduction in all possible risks. The result is, for example, better auditability.

I have recently heard of an excellent example of optimising power consumption in data centres. Using unsupervised learning (a method where a computer figures out for itself some relationships via a deep neural network without explicit instructions) to check the data centre’s workload, the computer was shutting off or switching on power supplies (computers, disk arrays etc.). Within a short period, it managed to find a stable model that did not affect the quality of service for clients, yet via targeted shutdown of unused sources or devices it reduced energy consumption by 30%, which is a particularly important fact and savings for data centres, as these have really enormous energy consumption. This is an interesting benefit.

How to optimise processes using data analytics?

As I said, all business activities nowadays leave a digital footprint. Take a look at the simple example of invoice delivery. You know that, when you have received an invoice, someone has scanned and forwarded it before. Each step has its timestamp. Based on the transaction logs, you can see where the invoice and other documents are moving, whether the trajectory is optimal or whether it can be improved. This is an area where data analytics is not yet fully utilised. In my opinion, this is a future direction that is worthwhile considering.

What are the trends in Data Analytics?

I’d like to mention attempts to use unstructured data, and there are only a few real business examples.

In addition, I’d point to sensible use of machine learning represented by the example of data centre power consumption control. However, machine learning is not suitable for many tasks because you are missing a description of causality. For a better understanding, let me give you an example of an insurance company that has some fraudulent cases in its stack of insurance claims.

You need to identify them, pass them on to investigators and decide in specific cases whether it’s fraud or not. These people need to know why you think it’s fraud. But the neural network won’t tell you. It just carries out calculations. It suggests the cases, but it doesn’t say it explicitly as it doesn’t know any facts that clearly show fraud. So follow the trends a bit carefully and apply them only where it makes sense.

It is worth mentioning self-service data analytics. Increasingly, businesses are striving to transfer capability and responsibility for data analysis to the end user, rather than leaving it to their IT department or external organisation. This has some benefits, typically the speed of processing, and information interpretation. I would call it the democratisation of Data & Analytics.

Another example is the Internet of Things (IoT). I might have a lot of sensors that travel around the country or the world in various ways, and I would concentrate their data in one place and extract them. A typical concept in this case is cloud processing, i.e. the trend of Data & Analytics in the cloud.

Crowdsourcing is also an interesting topic. I know a company that operates in the field of loans. It has developed credit risk models that tell it to whom it can lend and to whom it shouldn’t. These models have been refined over time. This company wanted to move on, so it used crowdsourcing.

They announced a world-wide competition, with payment of course, to find out who would come up with the model that best describes the situation of “bad payers”. The competition subscribers got access to anonymised data on those clients who paid and those who had a problem with payment. It really worked and they managed to improve the model.

What would you like to add at the end?

Technological tools are great, and there are many things to do in data analysis, but it really depends on the invention of people working with them in the company and their motivation.

I had experience of a project -built on an amazing data warehouse that covered virtually all the company’s data sources. The first two tasks were implemented – controlling reports and data reconciliation between the accounting and operational systems and some marketing stuff, and that was the end of the story.

The people who started the project were full of enthusiasm. Unfortunately when they left the company, the continuity was broken. It’s always a question of people. Technology and processes can only become truly useful when used wisely by the people.

Tomáš Mužík
Head of Delivery, CZ & SK

Related Posts

Interview About Data Analytics and Its Trends

Why it is useful to pay attention to data analytics and where its development might head in the future

Internet of Things and Industrial Analytics

Smaller firms can also benefit from IoT

Respectful Platform Modernization & Explainable AI

Perspectives from Ness’s CTO

Related Posts

Interview About Data Analytics and Its Trends

Why it is useful to pay attention to data analytics and where its development might head in the future

Internet of Things and Industrial Analytics

Smaller firms can also benefit from IoT

Respectful Platform Modernization & Explainable AI

Perspectives from Ness’s CTO