First look: IBM's Watson Analytics comes to the iPad
- 17 December, 2016 07:48
IBM today announced Watson Analytics Mobile for iPad -- an app that can be used along with a free personal or paid enterprise Watson account.
The app can import data from a cloud-based Watson account into a spreadsheet or CSV format, as well as use apps such as OneDrive and Dropbox to import files.
In addition, Twitter hashtags can be entered directly into the app for analysis.
Most features are available to free-account users -- except for analyzing Twitter hashtags, although if you're new to Watson, you'll get a 30-day trial of that.
The app "is not a replacement for the Watson Analytics web application," according to an e-mail from Marc Altshuller, general manager of business analytics at IBM Analytics. "As such, the mobile app focuses primarily on the discovery capabilities."
Those include answering ad-hoc questions, using natural language queries to discover patterns in data, and asking questions with voice commands.
In a quick test of the new app, it was clear that Web Watson has considerably more power. Like many mobile apps extracting useful information from data, Watson Analytics Mobile for iPad performed well with sample sales data that had well-understood values such as revenues, expenses, and countries.
Some data prep might be useful for other types of data before expecting easy natural-language answers to queries, though.
For example, I loaded in a spreadsheet of Massachusetts early-voting data from the Secretary of State's office, hoping to get some sort of histogram or bar chart showing distribution of early voters, but the bar chart needed some manual fiddling in order to get something close to what I wanted.
And when I asked for the relationship between number of registered voters and percent of early voters, only 100 points were shown on the bubble chart.
There are more than 300 cities and towns in the state, and it wasn't clear whether those 100 were a representative sample. In the Web version, all points displayed and there were more choices to create a customized data visualization.
One task the app is clearly designed for is monitoring Twitter hashtag sentiments on the go. For fun, I imported a few hash tags about my favorite NFL team, the New York Giants, to check sentiment about the team. There were few "ambivalent" tweets and a lot of positive ones.
Not surprisingly, tweets spiked late Sunday, December 11 into the early morning of December 12, when the Giants played a nationally televised evening game and beat the Cowboys 11-1.
(The game was the highest rated primetime regular season game in almost three years, excluding kickoff weekend, so Twitter activity was probably unusually high.)
The app can be downloaded from the Apple App Store.