Big-data projects are often ambitious, short-timed and plagued by talent crunch. Interesting thing to note here, big-data projects have a tendency to provide a sustained ROI. What makes it weird is that, if these projects are not baked properly, its negative impact will also contribute to the ROI, costing businesses time, money and flawed analytics.
So, what should a data scientist do? Learn from how successful startups are raised and put that knowledge to use for their organization. Following are the 5 important areas where big-data scientist could learn from startups:
1. Validate your model first and early:
Sure, you are tasked to define how you are going to lay your foundations for the data. How the data architecture will be laid out and how it is going to be used by people. No, don’t get tempted to start long road journey to development. Get your architecture, used cases and workflows validated as early as possible. Like any good product development strategy for cash strapped startups, nothing works as efficiently as the art of failing fast.
2. Don’t run after making your solution unique but address problem in generic way
We all have tendency to build world’s most sophisticated and state of art system. But, it is important to understand that a good platform is not just a world’s best cooked product but world’s most edible product. So, build your structure, stories and used cases from the problem that exist today, so that people could relate to it. Just like a startup, swimming in new territories is scary as unknown supersedes known, same applies for big-data projects. Get yourself a known used case which is faster and easier to build. If there is an essence of uniqueness, you will get there eventually.
3. No need to knit the whole thing, start with glue and get it going:
Yes, another problem that most of us face is to present the best thing out there. Cook the whole project in-house. This is costly ordeal that mostly fails as many unplanned exceptions starts showing up as soon as we start seeing our solution in action. So, against our conventional wisdom of doing it everything ourselves, we should leverage what is out there and build our prototype proof of concept with glued up solution. Once it is proven and all workflow and used case kinks are fixed, we could get that baked in-house.
4. Recycle is better if not best:
Building reusable modules, processes, data sets and architecture is crucial to business. Reusability leads to improved ROI on the effort required to revisit the logic again. So, it is important to strike a balance between doing 100% accurate, customized and unique solutions to more generic solutions with slightly less accuracy but more usable code that will ultimately save company’s time, money and effort. So, recycling must be embraced just like in startups where getting into most trusted and recommended tools is always first line of attack for achieving maximized ROI on efforts.
5. Don’t choose the most tedious path first:
Just like most of successful startups, progressive steps are often not taken from the tedious ways but from the simple ones. Similarly, your big-data project should learn this. You don’t have to create or choose the most difficult path as progression to your big-data initiative. If you come up with the most ambitious implementation to something, take a step back and rethink with open mindset if anything exists that gets you across with relatively less effort. A true data scientist is not always about solving data problem, but solving it in a commercially viable way.
6. Don’t run after the money, but focus on sustained value creation
Now, the previous point must be confusing you. No, commercial does not always have to be aligned with money, it could be a value driven system as well. Like any good solution, data architects should make sure to provide a model or methodology that facilitates sustained value creation and isn’t just a run for the money. Remember, many times data driven projects does not provide a clear financial incentives, and the benefits could vary across a wide range. So, let us focus on value creation in our implementation, because value ultimately translates to better business.
7. If it ain’t easy to explain, it’s still not ready for prime time:
You must have heard this about as many times as you don’t want to hear it. But it is important. If we can’t explain something in simple terms, it clearly speaks about how much we understood it. So, try working on getting to simpler side of things. This could very well be used as an indicator of which way to pursue. Often times, simpler ways are easier to execute and has higher tendency to succeed compared to ambitious and vague looking projects.
So, to conclude, for your big-data projects finds ways that creates sustained value with cost effective ways using quickly glued toolkits to prove the concept with simple and reusable science.
Rasalkhaimah, ras, al, khaimah, dubai, university, salford, manchester, @hishamsafadi, hisham, safadi, European, medical, center, business, entrepreneur, startup, economy, money, motivation, education, Leadership, Transactional, analysis, emotional, intelligence, organisations, development, innovative, technology, care, health, investor, investment, production, shark, tank, sharktank, USA, UK, London, group, european, canada, india, china, japan, KSA, projectmanagement, datascience, bigdata, IOT, internetofthings, cloud