Today we welcome the next generation of the PayScale Research Center to our site, or as we’ve been calling it: RC:TNG*. We’re generating more pages that are more specific and give you a better picture about the jobs you’re interested in researching.
PayScale’s consumer product development efforts work in three week intervals, in a development ideology that we call “maximally hobo.” It’s very close to the lean startup minimally viable product, except we think ours is more fun. We’ve spent the last three weeks hard at work building a new type of Research Center. I’ll spare you all of the technical details, but I’ll give you some links to read later on if you’re interested.
Basically, what RC:TNG does, is find the most interesting combinations of Job Title, Location, Skills, Employer, Industry, Degree, School, etc. and builds a page with salary information for that interesting combination.
Example:
- Salary for an entry-level software engineer with Java skills working in London.
- Entry-Level Personal Assistant Salary in London
- Graphic Designer Salary with Adobe Illustrator, Adobe InDesign and Adobe Photoshop skills
In order to complete this in three weeks, we had to impose limits on ourselves. One guideline was to say, “We’re only going to launch this for one country: United Kingdom.” As we continue to develop this product through the next three weeks, we’ll be expanding to more countries.
One of the other ways we were able to complete these changes in three weeks was by focusing on the big stuff. We found the interesting combinations, we built linking back and forth between pages, we generated the salary data, etc. But, not all the details are perfect. We would love your feedback. Just leave a comment here (or tweet at me, or email me)
*This should probably be “Research Center: Deep Salary Nine.” The current Research Center is actually the second iteration of the underlying data presented in the Research Center. But, let’s be honest, that doesn’t sound nearly as cool.
Further reading about machine learning techniques used for RC:TNG
- Unsupervised Machine Learning
- Association Rule Learning (a.k.a. Frequent Item Set Mining)
- Predictive Analytics
Also, if you’re interested in these things, you should take the Machine Learning class from Coursera. It’s really fun.
More from PayScale
Companies with the Most and Least Loyal Employees
Leave a Reply