Mako Metrics

Tableau and Lyza Together

I have struggled for years to turn raw data into useful information. The two primary challenges I face are preparing data to be analyzed, and actually analyzing it. I first heard about Tableau in September of 2008 via FlowingData, when they became a sponsor of the blog. I downloaded their trial, watched some of their on-demand training videos, and was blown away by how easy it seemed to allow someone to look at data from many different viewpoints quickly, and their examples are amazing. When I first tried using it with my data, I had difficulty getting the results I was looking for; Tableau’s approach to data was very different from what I was familiar with. After attending a few of their live training sessions, I finally grasped the concept of structured data. As I began participating in the Tableau User Forums and reading everything I could get my hands on about quantitative analysis best practices, I took to Tableau’s philosophy, the “The Zen of Visual Analysis” and the “cycle of visual analysis” described as “the iterative process of asking questions, creating pictures, discovering new hypothesis, and foraging for data”, like a duck to water.

Juice Analytics has a series of posts on the “5 Phases of Data Analytics Maturation” (Part 1 & Part 2), there they describe the most developed as: “Good information tools are just like an experienced safari guide.” I feel Tableau lets me experience my data with exploration. When I open Tableau, I am preparing to go on a safari into my data with Tableau as my guide.

After working with Tableau, a question remained for me, how do I best prepare data for Tableau? Depending on how source data is structured, I have varying degrees of flexibility. Different types of graphs can either be available or can produce different visualizations depending on the structure of the source data. I feel Tableau works best with raw data, where each row is a distinct event or data point. Many data sets that cross my path are pre-aggregated into a crosstab tables, and this type of data structure has limited flexibility. Tableau published an Excel Add-In, found on their forum, that un-pivots tables in Excel. It is a useful tool, and has saved me a countless amount of time in preparing data for Tableau. Another situation is Tableau aggregates data to one level of detail per worksheet. If I wanted to show a yearly aggregate and a monthly aggregate at the same time from a per-event data set, I would have to use a dashboard. This does not limit my ability to analyze the data, but there are times when I would like to show both on the same sheet.

Lyza is my solution for aggregating data at multiple levels and additional data preparation tasks. With Lyza, I can use different types of data sources (Tableau can join data sources of the same type, and multiple sheets in Excel very easily), and perform a series of simple queries to accomplish a complex task. By connecting Lyza to my data first, I feel that I have no limits in visualizing the story in Tableau. I can see Lyza’s goal to be a one-stop-shop for connecting to data, reshaping and visualization, while currently, the charting capabilities of Lyza are still maturing.

Lyza takes the process of querying data, and turns it into four simple possible steps. Each available step is intuitive and a joy to use. Like Tableau is my guide in visual analysis, Lyza is my guide in preparing data. I don’t remember when I first heard of Lyza, likely via their Twitter account, but I delayed trying it out because I did not immediately see how it could help me in my process flow with data. I tried out another application some time later with a complex user interface and was obviously over kill for my needs, but it worked with data in a series of queries, and it reminded me of Lyza. So I downloaded a trial of Lyza, and after watching a short video, I was up and running with my data. On my first day of using Lyza, I recreated a process in less than an hour; it previously took me a day of programming to originally develop. Instead of writing SQL and code from scratch, I was dragging and dropping with joy. Working with data in Lyza is like grocery shopping for me, I pick what I want and what flavor I want it in, I enjoy the selection of different variations of each data point available to me, and all I have to do is pick it up, drop it, and I’ve got it. Taking this analogy further, working with data in Tableau is like cooking a meal or a great looking desert (ah, the joy of cooking) with the items I picked out in Lyza. As I am cooking up a great visualization, if I realize I need some other data point, I go back to Lyza and drop it into the data set I have Tableau connected to, and voilà, I’m visualizing data as easy as pie (but hardly ever a pie chart).

The initial thing that I had to learn about Lyza was their use of color and what each meant. Once I did, their use of color became a great asset in easily seeing how things are set, where they came from, are going, and what can be done. I have made some massive series of steps, and when I find that I did something wrong along the way; Lyza provides two things that ease the process of finding out what I did. First there is the visual flow chart of parent and child steps, and second, there is a way to trace where columns of data came from, by reverse. Because each query step in Lyza is simple, I find myself creating patterns of steps to accomplish results. The act of splitting a complex query into its most basic parts is not an issue because this provides me a more elemental view of how my data is flowing, and makes it easier to troubleshoot and change.

With Lyza, I feel that I have complete flexibility in reshaping my data any way I want. Lyza has a previous function and summary functions that are really powerful, I will go into detail on these in a future post, and it is truly amazing how quickly Lyza allows me to get my data shaped the way I want it.

Both of these applications have allowed me to accomplish things that I previously deemed too complex to be done with a reasonable amount of time and effort, into tasks that are a joy and easy.

Thank you for reading, and I hope you return, or add the feed to your RSS reader, for future posts as I explore data with Lyza and Tableau together.

Disclaimer: as of this post publish data, I had no ties with Tableau, and a referral agreement with Lyza. I received no compensation for this article.