Previous post introduced the idea that software could take raw data and convert it to a usable news article. My friend John Bredehoft introduced the idea to me.
I think it is a great way for creative visualization of raw data. Good way to help us understand a mass of numbers.
What does an auto-written article look like?
Here are a few examples I found. They are all on the Forbes website, where Narrative Science is credited as the author.
Forbes Earnings Preview: Best Buy:
Since December 29, 2011 Best Buy‘s (BBY) stock has risen 16.5% to close at $26.93 on March 27, 2012. Best Buy looks to keep the momentum going when it reports fourth quarter earnings on Thursday, March 29, 2012.
What to Expect:
The Wall Street consensus is $2.16 per share, up 9.1% from a year ago when Best Buy reported earnings of $1.98 per share.
The consensus estimate hasn’t changed over the past month, but it’s down from three months ago when it was $2.18. For the fiscal year, analysts are projecting earnings of $3.37 per share.
Forbes Earnings Preview: Paychex:
Paychex (PAYX) reports its third quarter earnings on Wednesday, March 28, 2012.
What to Expect:
The Wall Street consensus is 37 cents per share, up 2.8% from a year ago when Paychex reported earnings of 36 cents per share.
For the fiscal year, analysts are projecting earnings of $1.51 per share.
Revenue is projected to be 9.5% above the year-earlier total of $519.6 million at $568.9 million for the quarter. For the year, revenue is projected to come in at $2.24 billion.
Both articles go on for several more paragraphs. There are other articles if you are interested.
Do those articles look like they were written by a computer?
I don’t think so.
Narrative Science has built into their software the ability to figure out trends and magnitude of changes.
The Smithsonian article I mentioned earlier says:
The software, for example, can interpret box scores to determine an appropriate angle for a game recap, distinguishing between a blowout, a come from behind victory, or a close loss.
At the end of his post Why do we have so little news coverage?, John gives an illustration of how this could have lots of value. He says:
For the last several years, last.fm and I have been compiling a list of the tracks to which I have been listening. Now perhaps this is not newsworthy (how many times did John listen to that song in a row?), but perhaps it could be formed into a story. Tweekly.fm tries to do this:
My Top 3 #lastfm Artists: Orion Rigel Dommisse (57), Zero 7 (46) & Client (32)
That’s interesting but it doesn’t provide a lot of perspective. That’s the same for most reports you get out of an accounting system or other database.
What if instead of that kind of raw data you could get a quick analysis like what John imagines:
John Bredehoft listened to the Orion Rigel Dommisse song “Skinwalkers” over 50 times last week, but also took the time to listen to some old favorites from Zero 7 and Client.
Apply that kind of visualization to a multi-year database of your customer or donor or vendor information.
Apply that to sales by product.
Apply that to statistics of each store’s sales by shift by department.
Look at the ministry efforts, inputs, or outcomes of each office, branch, or region.
That would be something valuable.
Consider this post I wrote – 4th State of the Plate report released for 2011 church giving trends. It discussed the trends in giving to local churches. My conclusion is that giving is less bad.
Applying the Narrative Science software to that information might give you most of the information I produced. I could have used their software to save me a fair amount of time, I could have their software write the first draft which I could then modify to throw in my opinions.
Next post discusses moving up the value chain, expanding on John’s response to my first post.
3 thoughts on “Words as a creative visualization? Part 2”
It’s intriguing to imagine how non-profits could use tools to visualize raw data. But some data lends itself to visualization better than others.
For example, I’m going to go out on a limb and predict that a number of churches will see a huge upsurge in attendance this Sunday (April 8). However, if I look at last year’s data, I will see a huge upsurge at this time of year, but it occurred much later in the month (April 24, 2011). Now we know why the dates vary from year to year, but how do we let Narrative Science know why the dates vary from year to year?