After two months of matching companies and job seekers, Louisiana Job Connection is updating the predictive algorithm that fuels our matching system. This is based on feedback from our users–more than 1,000 companies and 15,000 registered job seekers.
Employers and job seekers will likely see fewer or different matches when they log into Louisiana Job Connection. The matches they do see, however, will be stronger, more precise and more relevant to their specifications.
Based on users’ feedback and our experience, we determined it was time to tweak our system to provide better, more precise matches. Under our updated algorithm, skills and experience are still the top factor, but education is ranked higher than it was before. This means job seekers without the exact education required in a job will have lower match percentages. Highly skilled and experienced job seekers with the exact education specified by a company will have the highest matching percentages.
Additionally, job seekers’ regional preferences will now factor more heavily into the strength of a match. A job that is located outside of a person’s target location will have a lower overall matching score than it does now and will likely appear lower in the both the job seeker’s and the company’s matches.
Every predictive algorithm, including the one fueling our matching system, is a series of mathematic formulas set to produce results based on externally provided assumptions or factors.
When we launched in August, we told our algorithm to highly prioritize skills and experience and to assume that for the perfect job, most people would consider different locations, commute farther or even move. We still believe these things to be true. However, our users have suggested some refinements to meet their needs.
Based on feedback, we know a job posting’s education requirement is very important to many employers. And we also know that while many of our job seekers (almost 50 percent) did not set a regional preference for jobs, those job seekers that did pick a selected region (or regions) to target are very committed to that location preference and are not very interested in jobs elsewhere, no matter how amazing those jobs might be.
The Louisiana Job Connection system looks at a job seeker’s skills and experience and matches it to skills and experience required of each job posting active on the website. The system compares key phrases from a job seeker’s resume to the job posting. It’s smart and it learns. It knows that someone with “Microsoft Word” proficiency also has “word processing” proficiency. If a new computer programming language popped up tomorrow, our system would look at patterns, learn about the people who listed this as a new skill and improve its process. It’s that smart.
This is why we encourage job seekers and employers to be thorough with their postings. Because the more information the algorithm has, the more points it has on which to base potential matches.
Using the predictive algorithm, each job seeker is assigned a base matching score predicting their likely success in the position.
Here’s how it works: After establishing the base matching score based on skills and experience, our system now compares the job seeker’s education level and regional preferences to each job to assess how well the job seeker stacks up against the employer’s requirements—and vice versa.
A job seeker with the exact education level specified in a position keeps the base matching score assigned earlier. A job seeker without the exact education level specified loses percentage points based on the number of “levels” away from the exact education level they are. The number of points deducted ranges between 20 and 75, depending on the strength of the job seeker’s original match and how far “away” the job seeker is from the job’s requirements.
This means a job seeker with a strong base score (between 99.9 and 90) would lose fewer points for not having the exact education the employer requested than someone with a weaker overall starting base score. This is our matching system prioritizing highly skilled job seekers over those with less skills in a given area.
But our matching system knows that being the most qualified doesn’t mean a job seeker will take job if it is offered.
So the next issue our system considers is how likely the job seeker would be to take the job. Having established the most qualified candidates based on skills, experience and education, our algorithm now overlays a job seeker’s own preferences. When job seekers register for Louisiana Job Connection, they select a region or regions where they would like to work.
Our system now matches the job’s location to the job seeker’s preferences. If the job is not in the same region, the job seeker’s matching score will be lowered again.
Yeah, it’s a lot of work. Our system calculates this score for each and every job seeker in the system with each and every active job. It takes fractions of a second to calculate each match. It also learns each time it makes a match. When job seekers take action on a job, applying for it or deleting it, the algorithm learns from the kind of job seeker and the kind of action that job seeker took.
The ultimate goal is to end up with a system that predicts the best candidate for a job. In Louisiana Job Connection’s view, the best candidate is the person with the most relevant skills, experience and education who will take a job if it is offered.
We think we’ve hit that mark more closely with our newly updated algorithm. We’re excited for our users to see their new matches. If you have questions or need help, log into your Louisiana Job Connection account and select “Need Help?” from the top navigation bar to submit a help ticket.