Getting started with Big Data Analytics
What is Big Data?
The buzzword “Big Data” was probably due to the internet of things, be it technology or human behavior. This caused an increase in the interest of data collected.
However, what does one means when one says "Big Data"?
Definition of 'big data' by Gartner (2012)
" Big data are high volume, high velocity, and/or high variety information assets that require new forms of processing to enable enhanced decision making insight discovery and process optimization."
What this means is big data doesn’t only refer to the data from social media.
Big data comprises volume (or amount) of data, variety of data which may be in structured or unstructured formats and speed which data are generated or delivered.
5 items to keep in mind when thinking of embarking on Big Data Analysis
1. Plan your Business Objectives
Remember to plan your business objectives, keeping it measurable, to help remind the team of the objectives when ploughing through your data.
Some examples of business objectives may be :
Healthcare sector
- Cut waiting time at hospitals
Banking/Finance sectors
- Increase P&Ls through cross selling of products
- Reduce/ prevention of fraudulent cases
Telco industry
- Customer retention
2. What are the business impact for these objectives?
For the amount of work/ man power involved, is the returns enough to justify ?
3. How much relevant data do you have?
Looking within your company, is there enough relevant data for you to analyse ? Besides Google Analytics (or similar) from your web and/or social media listening / analytics, are there data beyond these that may be useful to your business objectives? For example sales data, customer profiles, focus groups etc
4. Are the resources available?
What are the skill sets required? Are we looking for data scientists?
Manpower skill sets generally includes , a programmer (IT person), an analyst and a business and/or product person. If your company has a data scientist, that a bonus to the team.
Looking at the software and hardware, many companies has software to help with the number crunching and visualisation outputs. Some example of these softwares include (not limited to) IBM SPSS, SAS, Tableau. Naturally, as one will be crunching large amounts of data, the hardware must be able to support such activities.
5. Is management on board ?
I believe this is self explanatory. :)
4 challenges to keep in mind
When starting on something new, there will be challenges. Some examples include:
1. Is the request for analytics an afterthought?
The results may not be able to tell you a full story if the request for analytics becomes an afterthought and not part of the planning of a project. This is due to insufficient collection of relevant data.
2. What happens if the outcomes are negative?
When one goes through the data, the results may be positive or negative. It may show that a hypothesis is proven untrue and the company have been moving in the wrong direction. What are the company’s next steps?
3. How much of the data do we share with management?
What is important to share with management and what to share with the working levels?
4. Do you have enough resources?
Resources cost money. This could be in the form of man power or tools required to do the work. Is there enough budget for the project? Does the team have the correct skill sets to perform their roles?
Case Study of using Big Data to make a difference in the Health Industry
(ref: http://www.todayonline.com/singapore/big-data-making-great-difference-healthcare )
To end off, I would like to share a case study from the Health Industry on how they used big data to improve their work.
Summary of the case study:
Business objective:
To cut waiting time at hospitals
Data points analysed:
Duration of 3 years of data collected
- Who were the patients?
- How many patients?
- How frequent patient visit in day and time?
- How serious were the cases?
Insights and improvements:
From data analysis, the results enabled better manpower redistribution matching patient's arrival pattern.
- Greater shift overlap in emergency doctor roster during peak hours
- Translated to 24% lower median time to first patient consultation with more serious conditions
I'd also point out the fact that in putting together data sets from Big Data, you're also likely to discover other data sets that you didn't think of creating - and that can create new initiatives, business ideas, research areas to find out more about etc.