A lot of words get used in technology and after a little while, no one bothers to mention what the word means. That’s too bad when the definition of a word gets changed, but that’s not the case with analytics. I found out that analytics is not a new word. It was coined in the 16th century to describe trigonometry, which makes me even more surprised WordPress’ spell checker always puts a red line under it as a misspelled or unknown word. I had someone tell me recently that they really weren’t sure what it was supposed to mean.
Wikipedia says “Analytics is the discovery and communication of meaningful patterns in data“. That’s what as data professional doing when we provide data in a manner which answers questions, such as providing KPIs, machine learning algorithms, or visualization. It’s not enough to be the keepers of the data library, data should also be used to provide meaning. Keeping that in mind, businesses all over the world tend to look for Adverity (adverity.com) or any similar company that has a skilled and experienced team of data analysts. Such service providers usually extract meaningful insights from available data and assist in the formulation of marketing strategies. Here’s another reason to work on analytics, the dollars the trade press is predicting will be spent on business analytics by 2018.
Steps to Providing Analytics
When describing the process for providing analytics, I am sure many people will recognize parts of the process as they are engaged in them now. The first step is to understand the data. Understanding the data does not only mean having knowledge of the structure of the data, as that obviously will be necessary to select it, but also needing to know how the business uses the data. Which fields contain the data they actually use? The second step is to preparing the data, including determining what data to include. Do you have all of the data you need to do the analysis? If the answer to that question is no, the analytic process will stop. You may have to exclude some data if it is incomplete or of dubious quality.
Once one has the needed data, it’s time to start the third step, data modeling. Modeling is where you categorize and make various decisions regarding the data. For example, if you are wearing a blue shirt and tan pants and you are looking at the laptops and you happen to be in Best Buy, you have found an employee. Determining if your model is evaluated in the next step. Generally speaking the analysis will include items where you know the outcome. For example, if you are trying to predict when your website volume will increase, you want to look at the historical events that made that happen. Marketing people do this to determine if the ad campaigns were successful, for example.
The Dynamic Analytical Process
After the model is created and sucessfully tested and evaluated, it’s time to deploy it and monitor the outcomes. One thing to remember about complex analytical models is they will probably change. One example of this is an analytical model many people are familiar with, the FICO score. FICO scores were created to predict credit risk. They have been tweaked quite a lot as the latest real estate crash showed that the fact a high FICO score showing someone paid credit cards on time was a lousy predictor of whether or not that same person would default on a mortgage. Netflix changes the movies they recommend when new movies come out. Things change all the time, so working on analytics means the work is never “done”. All the better for those of us who enjoy data analytics.
Data aficionado et SQL Raconteur