BD

The phrases Big Data, Predictive Analytics and Machine Learning have become all pervasive at conferences, boardroom meetings and even at water coolers. Everyone has jumped on the band wagon to take their business to the next level; Data Scientists are being hired at the rate of knots, Chief Data Officers have become commonplace and new technologies are being mastered and implemented, no stone has been left unturned.

However what many companies are yet to understand is this one vital component of Big Data success- Innovation! Without an innovative mindset no Analytics or Machine Learning endeavor will be successful. For better or worse Big Data is here to stay and to make it yield to your command, one has no choice but to be innovative, which is why it is essential companies start adopting a Big Data Inno-lytics (Innovation + Analytics) approach. Think of companies like Netflix, Amazon and Google, not only do they have technically gifted personnel they also have highly innovative, sometimes to the point of disruptive, people who keep pushing boundaries. These companies have changed the way business is done and are truly trend setters.

Netflix organizing a competition to improve upon their recommendation system is a case in point in Inno-lytics! The example given is innovation on a grand scale but fear not, not all innovation has to be larger than life, even the little things add up over time to bring about significant breakthroughs. Like everything else in business, innovation too needs to be scaled up over a period of time.

For example, say you have zip/post-code data available while building a customer churn model; one can be innovative by browsing the web for freely available socio-economic data, like average credit score/price of houses, pertaining to each of the area codes. This can then be included in the model to further enhance the overall predictive capability.

Deriving a variable from a group of others which is more closely correlated to the target variable is also another example of being innovative. A very simple example would be deriving and using Avg. Revenue per transaction instead of revenue and number of transactions individually in the model.

Organizations can also be innovative in the way an algorithm can be put to different uses, for example using Product Affinity or Market Basket Analysis traditionally used in retail, to determine part repair/replacement at automobile service centers dependent upon car type, age of vehicle and terrain (notice determining which factors to consider like car type also requires one to be, you guessed it – Innovative!). Further the same analysis can be used in health care to determine highly correlated treatments a patient will need to undergo dependent on illness/disease. Insurers can use this information to draft highly customized plans with appropriate premiums for their clients while also curbing inflated costs associated with unnecessary treatment provided by hospitals.

So the question is how does an organization promote a culture of innovation in their team of Data Scientists? A good starting point is to challenge them to publish an analysis in an area of their liking like sports, entertainment and travel to name a few. Borrowing from the practice at Google put this under the 20% initiative, that is, each member gets to work on his/her pet project for 20% of their working hours. this works out to 8 hours per week or a little over an hour and a half per day per employee. Besides challenging team members to think out of the box, you may stumble upon a new area of business.

An example here would reinforce all the point made so far. I love cooking and am an avid fan of Master Chef Australia. As a side project on my own time, I am building models which can predict the chances of a contestant winning based on factors like gender, age, types of cuisine they can cook, number of techniques they know, ingredients they are comfortable cooking with and ethnicity (this is an important factor given the fact Australia is a melting pot of many cultures). By tweaking this model I can use it to predict sporting outcomes and voila! With the right opportunities and a little bit of luck we can start an online betting site or sell the model to an existing firm for an astronomical figure!

Doing Analytics is not as hard as it sounds; it entails a structured empirical approach to derive insights through applied statistics and machine learning algorithms, using predominantly the left side of the brain. By throwing into the mix, the right side of the brain which is associated with creativity and innovation, organizations are setting themselves up for groundbreaking and exciting insights which will set them apart from the competition.


Related Articles:
Going Back to Basics. Understanding Big Data. 
How to Become a Superhero Data Analyst and get that Edge in the Workplace? 
The Collection and Collation of Data: Tips for the Business Analyst
 

Interested in a career in Data Science?
To learn more about Jigsaw’s Data Science with SAS Course – click here.
To learn more about Jigsaw’s Data Science with R Course – click here.
To learn more about Jigsaw’s Big Data Course – click here.
SHARE
share

Are you ready to build your own career?