What Data Analysis Really Is

I still have a message seating in my LinkedIn inbox from a contact who wants to know what data analysis really is. Every month I meet or get a call or receive an email from someone trying to know more about the relevance and true meaning of data analysis. So I thought it was time I gave a detailed answer everyone can access.

image: xlstat.com

Data analysis is the process of transforming raw data into useful information using standard analytical methods. 

An example we can all relate to at this current time in Nigeria is the voters registration. After INEC did the extensive voters registration in 2011, it had the raw data of the millions of people who registered. Now that's a lot of data. You don't expect that INEC would put all that records on Microsoft Word, get millions of A4 paper and print everything out. No. INEC had to get a data analyst (or a data analysis firm) who would take that huge raw data and transform it into much more useful information that will answer the following questions:

1. How many people registered in each state, each local government and each registration ward?
2. How many of all those who registered are male and how many are female?
3. At what rate of printing the permanent voters registration card will we be able to get everything ready before the 2015 elections?
4. What percentage of the people 18 years and above in each state registered (modelling it with population census data)?
5. How successful is the voters registration compared to previous voters registrations?
6. Can we see a visual representation of the states from the ones with the highest voting influence/power to the lowest voting power?
7. If people are expected to vote where they registered, how should we deploy our resources such that no place has too little or too much resources? And what margin of error should we prepare for?
8. If everyone who registered votes, how can we have the results computed and verified within the shortest possible time?
9. Using a modelling system, how can we optimize the registration process we used in 2011 to get a better process for the next registration exercise.
10. What is the estimated number of people who would turn 18 after the 2011 registration exercise and before the 2015 election?
11. What is the slowest rate of distribution of the voters card that will still let us meet the election deadline? 
12. How much did it cost us on average to register one person, print his voter's card and get it to him?
13. Can I have a dashboard with all these information? And can I view it on my iPad?

Those are the questions the INEC head should want to have answers to and only a data analyst can get him those answers. So now you have an idea of what data analysis is. It takes something as lengthy and boring as registration data of millions of people and turn it into a beautiful questions answering priestess.

One thing that confuses people is they think about data analysis in terms of a software. They ask me: so what data analysis software do you use? Isn't SPSS better than Excel? Can you use R? 

Data analysis is a science and not a software. Once you know what you want to do, then use whatever tool you can use (even manual on the paper with pencil calculations). Document your methodology. Give the error specifications. And build a sound analysis with a logical conclusion. Also show them you are not just some number churning robot, give them a business impacting recommendation. And don't forget to hand-over a beautiful dashboard.

Now you know what data analysis really is.