How You Can Do Data Analysis
Intro: If you read my last article What’s Involved in Data Analysis, by now you must have a pretty good idea what it is all about. With so much data flying around, we cannot pass the priceless opportunity of digging into them and getting all the possible information we would otherwise have missed.
On this one, I will discuss the various forms data analytics can take and where we can apply them. As usual, I will try to make this as painless as possible. The aim is to make it interesting enough.
There are a number of reasons why we might want to look into a body of data. Like the French scientist Claude Bernard famously said, “man can learn nothing except by going from the known to the unknown.” These are the ways we can use our known: data at hand, to get information we might not be aware we have.
Descriptive Analysis
This would be the most common of analysis out there. We do this on a daily basis in various forms. Every time you take a look at an object you are basically, even if only subconsciously, trying to ‘size it up’: capture the parts of it. This is so that you can match/compare it with things you already know. At the end of the day you call it a cat, a man, a Lexus RX model etc. Here you will occupy yourself with averages, standard deviations, percentage of wholes, high and low points etc.
In other words, descriptive analysis lists and summarizes the values of each variable in a data set. It helps you become familiar with a data set and to identify problems with the data. Like the name says, it aims to describe data content.
For sales data, you will identify high and low sales, totals and averages over periods, most sold items, highest buying customer etc.
Exploratory Analysis
This is sometimes called a diagnostic analysis. After the basics have been mastered in the descriptive approach, some projects are good to go. However, time and again this would need to be built upon for further insight. In exploratory data analysis you will check for relationships between parts of your data.
In essence, you are establishing a trends of sorts. As you pick an item or phenomenon in the data, you are isolating all factors surrounding it and attempting a diagnosis from these factors.
Continuing with our sales data example, this is where you try to to find correlations: match the bestselling products to locations, time periods and types of individuals. Items X and Y are bought mostly by people on the Lagos Island while no single buyer of item Z resides on the Island.
Prescriptive Analysis
You can also call this inferential analysis. Here you aim to find information from your set of data which will inform a specific recommendation. Given data at hand, what inferences can be drawn?
The recent ‘revelation’ about Google Trends data from Nigeria comes to mind here. It would prescribe entertainment for a surprisingly high predominance of youth, or at least a young at heart population. Similarly, this could come from a simple review of a customer preference poll on a set of products.
Predictive Analysis
I am usually hesitant to lump this type of analysis into a discussion of data analysis for beginners. It is data analysis, however, data at hand only provides a fodder for more (comparatively) advanced work. Predictive analysis involves predicting the future outcome of some indices in the data based on data at hand.
This is where business analytics gains popularity. It is always fascinating for management to know what products will sell well and those that will likely flop.
Sticking with our sales data, we try to determine how much we are likely to sell a particular product based on its sales performance in the past.
Outro: If you have started thinking of scenarios where you can grab that data and start looking for insight then I will invite you to the next article where I will introduce you to how to setup python for data analysis.
More resources from Principa and Career Trend.
Originally published at https://www.linkedin.com.
