Category Archives: Blog

amyjomartin followers

Activity of Followers for @jeffbullas and @AmyJoMartin

If one clusters together the Twitter followers for both @jeffbullas and @AmyJoMartin, one will find that Jeff’s least active followers tweet about 15 posts a month, while Amy has five groups that tweet below 1 post a month.

Back to the original FORBES article

Jeff Bullas followers

Back to the original FORBES article

Back to the article at LinkedIn

amyjomartin followers

Back to the original FORBES article

Back to the article at LinkedIn

Why President Obama Needs A Chief Data Scientist

During this year’s STRATA conference, President Obama introduced Dr. DJ Patil as his new Chief Data Scientist in a video message. DJ is a very well known data scientist and is even credited by some with coining the term “data science”. During his introduction of DJ, Obama said that he wanted to do a joke about Data Science but noted “half of the stuff my staff came up with was below average.”

Let’s decode this sentence into “stats speak” for a moment. What Obama meant was the median of the quality of those jokes was less than their mean. Thus the quality of the suggested jokes were skewed towards the end of bad quality, therefore he decided to drop the joke. That could be wrong, however, because all that was needed was one Joke. Thus if even all but one joke were terrible, it is the one joke that he could have used to start off his intro of DJ.

This omission is precisely why we need data scientists like DJ. We do not need all of the big data, we need the right data – and sometimes it is even only ONE dataset that we need. Even when most of the data within our Big Data cloud is ‘bad’ (aka. useless) we might be able to pull off a great prediction if we get the right dataset.

The same point is valid for models. We do not need many models (with perhaps the exception of a method called “random forest”), but rather we need only one model that is sufficient in its balance between accuracy and speed.

DJ, your knowledge and insights are needed. Let’s look for the right data within those 135 000 datasets that were made available to the public. We are looking forward to great changes based on data…

DJ - the first CDS of the US

Social Media Bots

Do Evil – The Business Of Social Media Bots

Warning: reading this might lead you to lose faith in the media and marketing world! Oh – you didn’t really trust them to begin with? Well, let’s dig right in and meet the now-removed Twitter account from @AI_AGW. He’ll hold an intelligent debate about global warming with you. But as smart as he is, you may be surprised to find that he is, in reality, a bot – a computer program. Bots like him cannot only talk to you, they can easily skew algorithms, influence your opinion – or, in some cases, cause a lot of trouble.



Before we dig into the dark world of bots, let’s find out what they are. Bots are algorithms acting in social media networks. But to the outside world, they look like a real user. They can come in all shapes and sizes, and they are borderline perfect. Some of them are very simple. And there are loads of services that will offer you bots, ranging from bots who will like whatever you post and fake followers to much more.

Building them is easy. You can actually try this at home. All you need is account, along with an RSS feed and maybe $10USD for 1,000 fake friends – all of them bots (Watch this movie to learn about @spotthebot and how he was built). Or better yet, download your very own bot software to wreak havoc on social networks in the comfort of your own home, within mere minutes. Much of this is even freeware (check out GitHub, for example). If you need a bot that can actually do conversations with you or others and pretend to be a human, you might check the Gonzales tutorial for code.

Social media bots can be scarily natural. A study showed that 30% of users can be deceived by a bot Tweet this. Well made bots can even gain your trust. For example, meet Lajello, a fictitious member in a book lovers’ network. He became the second most liked and appreciated person within this network. Why? Because he automatically recommended books to every other user like an Amazon recommender system. Lovely and friendly, right? Thus it should not be a surprise that 1 in 5 of us accepts unknown friend requests, openly letting bots into our world.


Bots are actually more common than you might think. Twitter is fighting bots via either legal actions or various machine-learning programs. Still, from time to time, researchers dig in and find that, for example, 7% of tweeps are non – human but spam bots. Companies like statuspeople () have built lists to actually spot fake Twitter followers, and you will be surprised to see how real some of them look. For example, 99.9% of @spotthebot’s followers are fake Tweet this… check them out. They are all bots!


So what is the problem with bots? An automated program like AI_AWG talking about global warming isn’t all that bad, right? Or what about the bot who tweets to you when your plant needs water? (Seriously – read about it here). In reality, bots can do things beyond our wildest dreams or nightmares. Here are some examples:


Can bots make you famous? Oh yes they can. Read here about how Chris Dessi (@ChrisDessi) created some fame with as little as 50k followers.


How about when bots persuade you to buy stuff? This is similar to email spam, or the nagging calls from the insurance companies to please buy their life insurance.


Let’s create bots that harm others; for example, our competitors. Think about what would have happened if 10,000,000,000 bots would have shown their identity shortly before the Facebook IPO. What would this have caused? (I actually know someone who tried to build such a system, except all of his bots got ‘killed’ shortly before the IPO. Facebook knows what’s at stake here.)

To harm others, even very simple bots can be useful. Just sign your worst opponent up with a lot of fake identities and help the world discover it. Newt GingrichMitt Romney and the German Conservative Party (CDU) are just some examples of cases where fake-following was proven or assumed to have happened.


One of the real impacts of bots is to skew public opinion. If you have built an army of bots that like, read, and engage on a set piece of content, you might as well influence what is trending. China tried this with the so called “5 Mao (50 cent) army“: over a quarter million bloggers who wrote articles for as much as 5 Mao per article to complement its government information politics. With my former startup (sold to WPP ) I had investigated this army for various clients and we saw a steady decline of their power. Perhaps they were replaced by bots? This is not as unlikely as you would think.


During the Arab Spring movement, we measured how the government was disrupting protesters’ activities with continuous tweets. By spamming the “stream” with tweets, important messages sent by activists were pushed lower on the page and out of sight by an automated system.

It is safe to conclude that the bot business has now gone beyond the scope of marketers  looking for fame and sales success. In fact, bots are now big government’s business. For example, the US Air Force revealed that it solicited Ntrepid, a California based company, to create software that would enable it to mass-produce bots for political purposes.


If there is one common denominator between these various uses for bots, it is that they all clone human activity in the limited world of social media. On a large enough scale, they might skew the ‘trending topics’ algorithms of various social media networks – and according to this survey, journalists often trust these trends. Think about the power if the tweets from your bot come up in google search results (now possible since google and Twitter signed a deal to make tweets more searchable). Thus the real power over the media is owned by the ones who run millions of social media bots.

Next week we will look into how one can spot bots and we will discover that some social media gurus have a lot of fake followers. If you cannot wait – just sign up for my newsletter or read chapter 6 in my book “Ask Measure Learn” by O’Reilly Media.

(republished from my Forbes Column)

The Stages of Analytics - Courtesy of Blue Yonder

Predictive Analytics – A Case For Private Equity?

[vc_row][vc_column width=”1/1″][vc_column_text]Framed Data raises $2M, 6Sense raises $12M, Reflektion raises $8M… The list goes on and on. What these companies have in common, aside from multi-million dollar investments, is that they are all in the market of predictive analytics. Predictive analytics is the art of making big data work by using past data to forecast future behavior. Who will churn? Who will buy what? Predictive analytics is at the core of Data Science.

Blue Yonder is one of those companies, and they just secured funding of $75 million from the global private equity firm Warburg Pincus . This is a unique deal in many respects. It was the biggest deal for a predictive analytics company in Europe and it was done by a PE firm, thus worth taking a second look – what is happening here?

I got an exclusive interview with Blue Yonder’s CEO Uwe Weiss (@WeissU). He explains that the market for predictive analytics is set for growth, how analytics meet transaction and why the future is in the automation of mass analytical decision in real time.

Read the full interview here or see below for the highlights and amazing insights of my interview with Uwe:

The Stages of Analytics - Courtesy of Blue Yonder

The Stages of Analytics – Courtesy of Blue Yonder

Analytics benefits the customer. 

Predictive analytics provide customers with a number of benefits resulting in significant cost savings and operational efficiencies primarily based on accurate real time forecasts which enable automated decision-making such as demand forecasting, dynamic pricing, replenishment, churn prediction, and predictive maintenance. Industries benefitting from predictive analytics not only cover retail and consumer but also increasingly finance, travel, energy and manufacturing.

Automated analytics is the future of predictive analytics.

Three different kinds of analytics have been defined by Gartner: descriptive, predictive, and prescriptive. Descriptive analytics describe the analytics that have been around since the dawn of business analytics in the 1980’s, traditionally known as reporting with simple descriptive tools, such as frequency distributions, charts and graphs. Predictive analytics use models describing past data to predict future trends. Prescriptive analytics provide recommendations to front-line workers. Uwe believes that there is a fourth category, “automated analytics”. Tom Davenport recently postulated that automated analytics is a further extension to prescriptive analytics, eliminating the need for the human to action prescriptive analytics. So, instead of a middle man changing prices or determining what kind of marketing email to send to their client base, it can be done through applications. Uwe believes that “ 99 percent of automated business decisions can be automated . If you automate business decisions, you need predictive analytics.” (read quote)

The predictive analytics market is established and growing.

The market for predictive data is forecast to grow at a compound annual growth rate of 34% from 2012 to 2017 reaching $48 billion, according to Gartner. Venture capitalists were quick to make early-stage investments into predictive analytics offerings. Uwe believes that any new startups in this space have to show how they are different in order to break into this already growth market. “The technology is at the plateau of productivity. People can use this technology now and produce ROI.” (read quote) Uwe believes that this is because the need for predictive analytics is fairly independent from traditional economic cycles. This view is matched by Gartner. Already in 2012 they predicted that predictive analytics technology is in a mature state. It is the lifeblood of a real-time enterprise and no longer an art in a “dark corner”. Working under these assumptions, it has become logical that Uwe sees the opportunity for a bigger enterprise software play. The time of predictive analytics being in a corner is over… Let’s go for scale.

Uwe Weiss @ Blue Yonder - Foto: Martin Leissl

Uwe Weiss @ Blue Yonder – Foto: Martin Leissl

Let’s go for scale! 

But hold on? Scale? So far, we see a lot of players offering niche expertise in one specific area such as churn management or marketing segmentation. Why? Because solutions are built around specific domain knowledge. The impact of domain knowledge in Data Science is often discussed; at the Strata conference in 2012, there was a lot of discussion over whether domain knowledge is important or can be solved by brute force. Senior experts in this domain such as Monica Rogati (@mrogati) and Xavier Amatriain (@xamat) from Netflix showed why domain knowledge is very important.

As Uwe explains, one can “build their software architecture around their domain.” The advantage is clear: “you can run very fast and catch a lot of market share” in doing this. (tweet this ) 

Uwe believes that is dangerous to remain in only one domain. “We go for the broader perspective and we have seen that we can – with our people and our engineering – tackle more domains and add domain knowledge relatively easily.” (read quote) He has geared Blue Yonder towards handling the bigger questions first – data digestion, time series management, and the handling of diverse machine learning algorithms. This generic-built predictive applications architecture allows Blue Yonder to integrate vague domain specialities into their platforms very rapidly. Thus, instead of software architecture being built around their domain, they are bringing domains into their software architecture.

Okay – I get it now. This is the new accord. This technology has reached the plateau of productivity . The market looks for an enterprise ready platform that can scale by incorporating domain knowledge, one domain at a time… And that’s why a private equity firm has invested in Blue Yonder and into the area of predictive analytics.

Predictive analytics is stable against economic cycles and will continue to grow fast.

Uwe points out one more reason why a PE firm would be looking to make such an investment. Predictive analytics is stable against economic cycles. “People buy food. People go to restaurants. People use trains and buy tickets, and so predictive analytics will have their space. The fundamental belief [of predictive analytics] and potential for explosive growth is responsible for private equity entering this area.” Even in an economic downturn, predictive analytics embedded into critical enterprise processes such as pricing, replenishment, logistics, and supply chain yield management will be cost-effective. Predictive analytics will always be needed and indeed the market opportunity here is huge.

When asked to comment on this investment, Joseph Schull, a Managing Director at Warburg Pincus, commented that  “the transformation of Big Data into actionable information is at the high growth frontier of enterprise IT .  Our investment in Blue Yonder is the result of an extended, international search for the opportunity to participate in this major new wave and growth opportunity.”  It is now clear that Warburg Pincus invested in their fundamental belief in the hyper-growth of this market.

With all of the excitement and development in the predictive analytics market, what’s next for Blue Yonder? What will the $75 million investment enable them to do?

“We want to place our bets on the right areas, so America is definitely the next step on our internationalization roadmap.” (read quote) Even with the geographic expansion, Blue Yonder will continue to expand on the enterprise software level. “Product-wise, we will focus on the domains we have defined for ourselves. This gives us a lot to do for the next 24 months. I am convinced that we will see a mass market for predictive applications from 2016 onwards. All of the global 2000 Leading Companies in all relevant industries will need to adapt and use it . ” (read quote)

Thus, the prediction about predictive analytics is that there is more to come. Stay tuned – and subscribe to my newsletter to learn more about Big Data and why we actually want Small Data! The full interview can be read here at

(republished from my original FORBES article)[/vc_column_text][/vc_column][/vc_row]