Like many of you, I am experiencing organizational change at my company in the new year. Alteryx has a new CEO (Mark Anderson, best known for his work at Palo Alto) which means I have a new boss and a new way of doing things. Naturally, this makes me wonder what this year will look like. I could gossip monger and make calls and try to figure this out in the standard ways, but hey… I’m a trained analyst. This is what data analysis is all about. Let’s see what a quick, structured analytics project can tell me about what it will be like working for my new leader.
Reading reviews of my new CEO’s old company on Glassdoor or Indeed would potentially tell me what I want to know, but there are two small hiccups with just pulling up the sites and browsing: (1) there are a lot of reviews to read; and (2) it’s likely I would suffer from confirmation bias and see only what I want to see in these reviews. Fortunately, I work at Alteryx and can easily apply some advanced analytics to see what the future holds.
On a Monday evening I pulled up Alteryx and had the answer to my question in 18 minutes. Yup, I timed it. My very rough workflow was straightforward and required me to scrape the Indeed website for the reviews, filter them to when Mark was the president, visualize the general idea with a word cloud, then run sentiment analysis on each review to see if they were positive or negative. These concepts sound time consuming or some may say difficult, but it truly was a breeze. Eighteen minutes, remember?
For those of you who may be unfamiliar, Alteryx is an advanced analytics platform designed to run analytics completely from end to end. As you can see with the above workflow, data was ingested, cleaned and analyzed all with one solution and without a lick of my own code (though I could have done this programmatically in Alteryx, if that were my preference).
Downloading the data was done by providing the URL of the webpage I wanted to scrape. Seriously – that’s it. The data returned was messy since some html markup came with the text I wanted to grab, but we have building blocks (those icons you see above) to get rid of the junk and keep the good stuff. As a bonus some reviews that were hidden on the view of the web page came through with my scrape. Check this out…
Hidden data! Love it!!!
The data is in. Time to start analyzing.
With 3 clicks, my word cloud was generated. I had to tell the software what field of data I wanted to analyze, add a view of the cloud and click run. With that – voilà, we have some results.
We could spend some time making this cloud prettier, but hey, we’re going for speed.
Now we have the general idea that the impression of my new leader is good. I like that “great” is the most frequently used word. However, I still want to know how many of the 93 reviews posted were negative and what they had to say. To do this we turn to sentiment analysis.
Prior to coming to Alteryx I had to use python for sentiment analysis. I am very happy to say that now I can perform this task with a few clicks. The configuration for the sentiment analysis building block is below. You will notice that I have to tell the computer which language use for the text analysis, which algorithm I want to apply (all behind the scenes of course – no coding necessary), and which field we are examining. Also quite helpful is the ability to tune the cutoff for what is considered negative or positive. As you likely know, people who post online tend to lean towards the negative so when analyzing survey data, reviews or other posts you might want to adjust this setting. In our case we are going to leave the default in place. @MarkAnderson is winning at life. He doesn’t need my help.
The moment of truth: the configuration is done, let’s check out the results.
Well folks, I think I hit the jackpot with this guy. The feedback is overwhelmingly positive. The first negative I dug into was negative because he was leaving. How could it get much better?
To be honest, I wasn’t expecting positive results simply due to the medium. Generally, people lean a bit negative online but this quick glimpse into Palo Alto back in the day is showing that Alteryx is still looking good! My nerves have been calmed with 18 minutes of work. Not to be overly cheesy, but I think I just experienced our tag line… “The Thrill of Solving.”
If you are interested in testing our analytics process automation platform sign up for a free trial here. We have plenty of people ready to help you with building out your first project. Reach out to me or anyone on the Alteryx community whenever you have a question or need a hand. Additionally, I’m always looking for interesting things to analyze. If you have any ideas of what you want to see, post them in the comments!