Data & Analytics Strategy

2016 Lessons Learned in Data Analytics

Since we launched TCB Analytics a little over one year ago, we’ve engaged in data analytics projects for clients across various industries. Some projects were focused on data visualization and reporting, others on data scraping and machine learning, and all of the projects required initial data cleansing.

Lessons Learned in 2016

The New Year gives us a chance to reflect on lessons learned so that we can improve and prepare for the future of data.

Automate, Automate, Automate!

This cannot be stressed enough. If you find yourself continuously doing the same task, whether it’s renaming or reordering columns, joining data sets, or even calculating various metrics on a regular basis, you should write a script to automate the job. It’s now 2017 – there’s no need to manually create charts or copy and paste in Excel anymore. We’ve created R scripts for specific tasks that can be easily modified to work with various data sets from our clients. For example, we receive data sets on a regular basis from a client that need to be QCed, joined with another data set, and visualized in a Tableau dashboard. Rather than manually update the Tableau dashboard every time the new data is received, we wrote scripts to automate the update process. You can read more about how to do that here.

Save Time with Templates

Many of our projects involve data cleansing, manipulation, visualization and dashboarding using Tableau and Shiny. Since most Shiny dashboards require a few filters, a table and at least one chart, we created Shiny project templates for these various use cases. This saves a lot of time in setting up a Shiny project. With pre-built templates, we can quickly create a basic dashboard after simply adding a client logo and modifying the data input.

Remember the End Game

It’s easy to get lost in a sea of data, especially if you are constantly analyzing the data yourself. Always keep the original question in mind so you stay on target. Ask yourself – why am I creating this new variable? Will it improve the model so that we can answer the question at hand? Is joining this new data set adding new context that’s relevant? Gut check yourself and get back on track if you find yourself going astray from the problem.

These three lessons learned have been huge time savers for us, allowing for more time for critical thinking and communicating our findings to clients.