Author: Anderson MacGyver

The Disaster

Stockholm 1628. It is warm and sunny on this day in August. Everyone who can, makes it to the piers to witness the historic event. Many even travelled for days to be part of the moment when the proud of the Swedish nation setsets sail. The Vasa is the most powerful navy vessel that the Swedish navy ever took into service. It has an unmatched firepower and might be the decisive factor in the raging war with Poland-Lithuania. Considering its costs, it better be! The 64 massive bronze cannons and the rich decoration with hundreds of painted and gilded sculptures cost a fortune and make the Vasa a huge asset to the Swedish Kingdom.

On this clear and sunny summer day, the Vasa sinks on her maiden voyage, 1300 meters after setting sail, when the first light gust of wind fills part of her sails. Some 30 people drown.

What Happened?

How could this moment, which was supposed to be triumphal, turn into a devastating catastrophe?

The answer is shockingly simple. The design of the ship was not suitable for the heavy weight of the large number of canons that set the ship’s centre of mass too high for the ship to be stable.

Another Golden Age

388 years later, history repeats itself once again. Just like during the days, the Dutch refer to as “the Golden Age”, we are living in the middle of a time of adventures and opportunities. The ships of the VOC (Verenigde Oost-Indische Compagnie) brought back not only precious goods from their journeys to distant lands, but also not to be underestimated insights on discovered cultures, flora and fauna as well as important naval knowledge.

Today, most companies find themselves surrounded by an ocean of data. Just like in the past, sailing this ocean can be an adventure leading to new fantastic business opportunities. In addition, new insights and trends can be obtained from the journey for the management to facilitate the right decision making. Another Golden Age! However, the journey is tricky and requires knowhow and skills. Just like for the Vasa it is easy to capsize and sink with missing knowhow and the wrong approaches.

Less Is More

There is another lesson to be learned here: less is sometimes more. Even for data analysis! While it is widely believed that more data gives you necessarily more and more reliable insights, this is in most cases a dangerous assumption and more often than not simply wrong! If you have sufficient large statistics, i.e. huge datasets, modern algorithms can create reliable insights even with polluted and incomplete data. But the vast majority of companies do not have enormous datasets like Google or Facebook. They do not analyse hundreds and thousands of Terabyte of data per day to deliver internet-search results or analyse their clients’ behaviour. Their datasets are a thousand and even million times smaller. Here, a good data quality is essential. ‘’Garbage in, garbage out” remains a valid statement for the vast majority of analyses. The most effective and efficient approach is usually a ‘light’ and transparent analysis that is easy to verify and understand, based on a few but powerful variables. Trying to ‘squeeze’ all your data and variables in a black box and leaning on the results is a risky enterprise and in most cases doomed to fail without prior knowledge and thorough understanding of the data. Less is sometimes more, the predictions of a predictive model, based on a machine-learning algorithm, for example, might be significantly worsened by the addition of irrelevant variables.

The Vasa might have played an important part in the war with a few cannons less on board.

Prevent yourself from the risk of sinking and ask for our advice on your (potential) data analysis.

…or: For Hard Results, You Need to Get Soft!

The Neglected Side of Big Data

“Big data is not about data! It is not even about technology!” – Quite a bold statement, especially if you follow the public debate on this topic. Most articles and discussions are centred on technical developments like Hadoop, cloud computing or self learning algorithms. This leads to the impression that big data projects are successfully accomplished by just finding the right technical tool. But this point of view falls utterly short of a fundamental characteristic of big data. You might call it the soft side of big data. This recklessly neglected side more often than not is the decisive factor that determines whether a big data project or programme succeeds, or fails.

So, what is this mysterious soft side of big data? In the style of the 3 V’s (Volume, Velocity and Variety) often used to define big data, you may summarize it as the 3 C’s: Creativity, Collaboration and Culture.

Creativity

At the beginning of each big data project stand creative ideas on how to take advantage of data. These ideas typically are formulated in questions that start with “What if we would/could…..?”. Think for example of a logistics company that could ask “What if we would combine our own data with open data, such as weather data or traffic data? Could we improve our daily delivery predictions and routing? And if so, by how much?”. But also during the data analysis, the core of each big data project, creativity is essential. With some creative ideas and programming, the analysis code can be optimized in performance and new and often unexpected insides can be revealed, like e.g. that an inefficient use of the operation theatre in a hospital is not caused by emergencies but by the long scheduled surgeries.

Working with big data involves a lot of experimentation, since most of the times you are exploring uncharted territory. You need to be creative to overcome and solve the challenges you are facing during a big data project and analysis, and creativity is an essential ingredient to succeed. Without creativity every big data project is doomed to fail already from the beginning.

Collaboration

To take full advantage of the possibilities that come with big data one needs the collaboration of multiple departments of a company. The focus of the analytics part of a big data project is provided by the business, commonly by sales, marketing or operations. The technical elements themselves are under the responsibility of IT or an autonomous data department. Essential for a fruitful collaboration is that the interdisciplinary members develop a common language. This is not trivial, seeing that most collaboration members typically come from very different backgrounds. A proven approach to swiftly support the development of a common language is through visualizations that act as a common basis of understanding. As part of our day-to-day business we often support the creation of a common language for business and IT using our proven operating model canvas. The same technique can be used to enhance the communication in big data projects.

Next to this internal collaboration, the collaboration with external partners such as research institutes, universities or consultants can help to deliver the desired value of a big data project.

Culture

Successful big data projects and programmes require a change in culture for many organizations. The related projects and programmes are very dynamic and the necessary approach is an agile one. A lot of experimentation is involved with iterative cycles. This requires a dynamic, creative and inspiring environment where people dare to try something new and unconventional, an environment often found at start-ups. This is why various large companies like ING approach the big data challenges by first founding, and later on integrating start-ups. However, also the creation of company internal, interdisciplinary teams for big data projects can be very successful.

It pays off to pay close attention to the 3 C’s of big data. It will not only support the project and help to reach the defined goals faster and more efficiently but it is also likely to create higher quality results with a higher impact. For hard results, you need to get soft.

Do your big data ambitions need more creativity, collaboration or a more inspiring cultural environment? Do not hesitate to contact Anderson MacGyver.

A Huge Chunk of Marble!

About 6 meters high, over 7 tons heavy and not of the best quality. A few, well-known artists of the time, the beginning of the 16h century, accepted contracts to turn this enormous chunk of marble into one of the twelve statues which were meant to decorate the buttresses of Florence’s cathedral. None succeeded. Finally, it was a 26 years old genius that turned this massive block of Carrara marble into one of the world’s most admired sculptures.

Data Science and the Fine Arts

On the first sight, data science and sculpturing have little in common. The former uses advanced computers and code to generate insights and knowledge while the other relies on hammer and chisel to create art. But viewed in a different light, quite intriguing similarities are revealed.

Faced with a large raw-database, the data scientist commences his work in a similar way to a sculptor who starts by carefully examining his chunk of marble. Possible cracks and the type and condition of the stone determine the tools and feasibility of the project. However, after this initial check, both data scientist and sculptor need to proceed with great care. Hidden, microscopic cracks in the stone can have similarly devastating effects as unnoticed biases in the dataset or small bugs in the analysis code. While the damage in case of the sculpture is clear to the eye, errors and bugs in a data analysis cause false results that are often hard to identify. It is all about experience, the right tools and the right ideas. “A man paints with his brain and not with his hands.” (Michelangelo Buonarotti). The same holds true for sculptors, I suppose.

Who Said Anything About Creation?

The results can often be surprising! “How could they possibly make this?!” is a question that comes to my mind whenever I admire a sculpture that is made of cold stone and yet looks so realistic, almost alive!

Also the results of an advanced data analysis can lead to astonishment. Newly discovered insights, as for example “three quarter of your customer profiles are wrong”, or the high accuracy of a devised predictive model, for example to predict the number of orders or delivered goods for the coming days, trigger questions like “How did you do this?”, “How could you create these insights?”.

But a data scientist does not create the insights. The ‘created’ insights have always been there, in the data, waiting to be revealed. A data scientist’s task is simply to provide the access.

It is said that when Michelangelo was asked how he was possibly able to create his ‘David’ out of this chunk of marble, he answered: “I did not have to create him. He was always there, in the stone. I just had to remove the marble around him.”

In the end, ‘David’ did not end up on top of Florence’s cathedral but found a more prominent place right in front of Palazzo Vecchio (and since 1873 in the Galleria dell’Accademia), to be admired by Florence’s citizens and tourists alike.