- Share this item with your network:
- Download

Business Information
- FeatureData visualization techniques, tools at core of advanced analytics
- FeatureFour steps to make a data visualization project succeed
- FeatureVisualized data aids in building, updating predictive models
- FeatureMark Curphey just gave you the tools to make safe apps -- for free?
- FeatureBig data architectures face big hurdles with technology integration
- FeatureBig data governance steps into IT spotlight as architectures expand
- FeatureThree steps to ponder for your first cloud-based analytics project
- OpinionSports data analytics isn't always a slam dunk
- FeatureMastering descriptive data analysis yields better predictions

pressmaster - Fotolia
Three steps to ponder for your first cloud-based analytics project
For the uninitiated, analytics can be overwhelming. How you approach your first analytics project in the cloud can make it bloom or lead to doom. To hit the big time, start small.
As with any data analytics project, cloud-based analytics requires asking the tough questions about data: How much do you capture? What do you keep and for how long? What do you filter out? For some, casting a wide net and filtering later is easier and worth the extra storage costs. For others, examining data for usefulness as it's presented may constitute a leaner alternative. Either way, great data does not guarantee great cloud analytics. Ponder these three points as you venture into your first analytics project in the cloud.
1. Focus on the business goal. In their book Competing on Analytics: The New Science of Winning, Thomas Davenport and Jeanne Harris stressed that analytics is a business initiative, not a technological one. Figuring out what you can extract from existing data is less likely to advance the business than specifying what the company needs to know and determining what data is required.
2. Choose carefully. "The more [data] you have, the less valuable it becomes," said Simon James, global performance analytics lead at SapientNitro. Unrestrained data growth is expensive and may result in new business strategies based on obsolete data. That's always true -- except when it's not. Long-term medical studies spanning multiple family generations must include vast historical data along with new information. The famous Framingham (Mass.) Heart Study, for example, began in 1948, 68 years ago.
3. Start small but provision adequately. For a pilot, choose a non-mission-critical project that can falter without dire consequences. Ensure that adequate resources -- staff and infrastructure -- are budgeted for and allocated. Stumbling because you attempted to do cloud-based analytics on the cheap is likely to reduce the odds of getting a second chance.
Next Steps
Latency is unacceptable in streaming analytics
IBM's Nagui Halim is the father of streaming analytics
Streaming analytics is simple, getting there isn't
Related Resources
- Experts Answer Pressing PaaS Cloud Development Questions –SearchBusinessAnalytics
- Focus: Cloud application development –ComputerWeekly.com
- Focus: PaaS and the next generation of cloud software –ComputerWeekly.com
Join the conversation
1 comment