Published on 2017-09-06 23:48 (LinkedIn: sethsanu)
created a Power BI report based on my LinkedIn profile data. LinkedIn members can download their data to CSV files via https://www.linkedin.com/psettings/member-data. The archival process takes about three days, with the first set of CSV files arriving almost instantaneously.
Custom Visuals. In the consulting world there's an inherent fear of using custom visuals, and rightfully so - some send data to third party sites, some have the reliability of Pontiac Aztek (similarly, the coolness factor of custom visuals is as subjective as a Pontiac Aztek). I only used my favorites ...
For the company images, I downloaded the LinkedIn versions and uploaded them to my OneDrive. From OneDrive there's an embed link where a public URL for each image can be obtained. An alternative would not be to host them, but to use hot links to existing company logos hosted on the company site. Problem is I've worked for a couple companies that have shut down - heaven only knows how LinkedIn has these images when a web search turns up empty.
Data Quality. In creating the report, I realized that my LinkedIn data wasn't up to date. In some cases I had jobs without start and end dates, and skills that were (nay, still are) missing. The Power BI report was a wake up call to update my profile.
Power BI publish to web. To publish to web you need a Power BI Pro License. I have one through my Office 365 subscription on Sanusi.org (a domain squatter in Korea has Sanusi.com ransomed for the next decade). Note that when you do this you're sharing all information with the world; so heed the warning messages for the privacy conscious.
In the consulting world, our resumes include a list of skills and a numerical proficiency rating. I adopted the same design using the skills list that LinkedIn provided. I added a column for proficiency and start / end dates. Primarily so I can infer the skills to the job I was in, but also because I wanted to downplay skills I no longer wanted to emphasize (too many recruiters calling for ABAP and MSAccess ... not like there's anything wrong with those, they're just not my cup of tea). Ah data gaming at it's finest.
The trickiest part was connecting start and end dates to a date slicer. I've done this on past projects and implemented the same technique: cross join a date table against the summarized start / end date table, finally filter out only the intersections. This allows the date slicer to show only the jobs I was working within the slicer period. I'm sure there's a better approach and I welcome comments / suggestions / philosophical entreaties.
This is V1, I hope to refine it a bit with the 2nd batch of the LinkedIn archive data (my history of likes). I also would like to templatize it as a pbit file so others could use it along with their own LinkedIn data export (provided they fill in the data model blanks).