Intelligent data discovery has amazing potential to alter today’s manual approaches to data visualization. Automating analytics with next-generation search and data discovery provides rapid insights powered by advanced analytics and artificial intelligence. Over the past few years, several vendors introduced smart data discovery concepts. Mainstream offerings are also slowly adding them. Today we are seeing an exciting new generation of savvy analytics vendors entering the market. One of these up and coming players is Tellius.
Tellius – Tomorrow’s Insights Today
Recently I was contacted by Tellius to review a demo. Apparently they read my sassy DOMO blog and wanted my unfiltered, honest feedback on what they were building. If you are a regular reader of my blog, you likely know that there are literally hundreds of data visualization vendors. Many of them appear to be commodity offerings on the surface. To fully appreciate what makes a visual analytics solution unique, you need to experience using the solution by analyzing your own data in a hands-on manner.
After running through a standard informational industry analyst briefing, I requested a hands-on evaluation. When I wrapped up my testing on Tellius, I came to the conclusion that they have a pretty cool smart data discovery offering for automating analytics. They are someone to keep an eye on as they continue to evolve.
Tellius provides seamless unification of visual data intelligence, actionable insights and advanced machine learning capabilities.
My Top Takeaways
In exploring the Tellius offering, I absolutely loved the ranked Smart Insights. Smart Insights reminded me of when I first saw BeyondCore, a smart data discovery solution that was purchased by Salesforce last year. I also liked Tellius guided search and full predictive life-cycle capabilities for operationalizing models. Other nice features included point-and-click data preparation, calculations, out-of-the-box time intelligence, visual themes and “Data Fusion” aka data blending.
To be balanced in my review, I noted room for Tellius to refine user experience, increase the array of available data sources and add native mobile apps. All in all, for a new offering Tellius brings a lot of functionality. Here is a brief peek into the current version of Tellius that I tested.
Connect, Explore and Search
After logging in, I was able to easily connect to a variety of data sources including files, popular relational databases, big data solutions, cloud apps and cloud data warehouses.
When I connected to data, Tellius provided me automatic data profiling and cleansing options. The solution also allowed me to define values for common discrepancies such as mean, zero, and custom value. Transformations that I noted include adding or renaming columns, splitting, merging, filtering or moving columns. There are also set target value, find and replace functions.
Now that my data was prepared, I was provided several options for analysis.
- Create Stories
- Visualize my data
- Create a Predictive Model
- Persist to a Database
I initially decide to visually explore my data. Simply by clicking on an add visualization button, I was able to get a list of intelligently ranked Recommended Charts of key discovered unbiased insights and explanations expediting my analysis.
Smart Insights might just be my favorite feature – I loved it.
Tellius In-Memory Compute Engine (ICE) Insights automated the data crunching and insight identification process to provide that list of Recommended Charts. It also automatically created an entire Viz Library of variable combinations for me. Automating analytics with Tellius was simple, fast and fun.
Smart Insights finds segments in the data that are not initially labeled in the data set while finding optimal discoveries. These slices of data are calculated using machine learning algorithms. An example segment might be people with a salary between $30K to $38K, age 25 to 35, who live in California. The insight on that segment is this group has a maximum probability of sharing an apartment with a friend.
As I added visualizations to my Vizpad, I noted that I was able to choose a color theme and visualization type. I also could contextually filter all visualizations on a Vizpad. In my opinion, contextual filtering is a must-have feature when I review visual analytics solutions.
Today Tellius provides the following chart types.
- Bar, Line and Combination Charts
- Pie Charts and Area Charts
- Scatter Plots
- Polar Charts
- Heat Maps
- Radial Gauges
- Word Clouds
- Simple Numeric Cards
- Geospatial Maps
I found sizing and placing charts on the Vizpad canvas to be easy to do after I learned and adapted to the Tellius approach of changing settings in the top right menu of a visual versus dragging and dropping fields.
Continuing on in my evaluation, I then tested Search. Tellius offers two flavors of Search – Guided and Natural. The Guided Search prompts the user with field names and possible combinations to explore in the search area. Natural search infers Natural Language Processing. Notably according to a recent Gartner article, Natural Language Search is expected to be in most modern analytics solutions by 2018.
Using Tellius Guided Search, I was able to get instant answers and suggestions for ad-hoc questions across multiple data sources via the “Data Fusion” engine. An auto-suggest feature was helpful for filtering through many available tables and columns. I liked that I did not have to pre-calculate the data or create a model. Tellius can run ad-hoc query on hundreds of millions of records with sub-second response times.
Another relevant Search feature that I noted was transparent result permissions. Search recommendations honored data set permissions. Users only get results and see recommendations based on assigned security roles for data access and governance.
Easy Predictive Analytics
Next I looked at the embedded predictive modeling and evaluation functionality. I do like that data scientists and business users alike can use this feature. Realistically I could see data scientists creating a model and easily sharing it with the non-technical audience on the same platform. Those functionalities are often limited to silo data scientist tools and require business analysts to copy/paste R code or try to call a REST API from a data visualization tool. Having the end-to-end capability within Tellius was a pleasant perk.
Tellius predictive algorithm types included ranked feature selection along with several time series / forecasting techniques (ETS, ARIMA and STLM regression), Clustering (KMEANS) and Classification (Decision Tree, Random Forest and GBT).
After a predictive model is created, Tellius provides a simple point-and-click option to use that model on a new data set. For instance, I could apply my bike buyer prospect list to the classification model to visualize who in that list was projected as a likely bike buyer and add that to my dashboard. Making predictive models operational is awesome for the business to move from “so what” to “now what” actionable analytics.
I found Tellius to be a refreshing, cool newcomer. Often when new data visualization vendors brief me, I find it difficult to understand what makes them truly unique and beautiful in today’s overly crowded analytics market. Rarely, if ever, do I see innovative smart data discovery vendors. When I do find them, I am always delighted to see what they have cooked up and how they will revolutionize the industry that I know and love so much.
What I truly adored about Tellius is their unified approach to automating analytics. They combined analytics automation, visual data discovery and intelligent analytics with additional features to operationalize data science algorithms. They had foresight from the start to enable the use of custom predictive models to provide strategic analytics advantage. Essentially, Tellius brings intelligent analytics to the masses.
If you are interested in learning more about automating analytics with Tellius, please review the following resources.