Nearly every job in America, mapped in detail | Hacker News Reader

Nearly every job in America, mapped in detail

washingtonpost.com

132 pointssebg45 comments

This it not the TYPE of job, but by INDUSTRY it appears from looking at the LED (Longitudinal Employer-Household Dynamics) description.

The difference is incredibly important. Have 500 workers including janitorial, secretarial, programmers, and executives at a high-tech company and they are all reported as high-tech workers regardless if the highest tech they touch is a floor waxer. Or have some IT people at a hotel and they are classified as hospitality.

I never like these industry classifications because they aren't actually very helpful in looking at employent statistics except in aggregate.

Your observation is spot-on and I agree that it's important not to misread this data set, but I think it still has some valuable uses. While I would rather see more detail, I am interested to see how different industries impact overall employment in the economy even if I can't bucket all the individual jobs under one heading.

With that said, given that a lot of companies contract out for non-core work (e.g. janitorial, kitchen/food-preparation in cafeterias, etc), I wonder how that is handled in this dataset? In particular, what location would be associated with a given contract worker. Would it be their employer's address or the address where they provide services?

Employer. These are all done through payroll numbers and some self-reports to fill in the gaps maybe. I'm guessing, similar to the monthly ESR establishment survey.

Direct link: http://robertmanduca.com/projects/jobs.html

The data is available in (gzipped) CSV here: http://lehd.ces.census.gov/data/

And LEHD also has its own visualizations here: http://lehd.ces.census.gov/applications/

Great map! I notice a large amount of green. In fact, most major cities are completely green. Could it be an even tight mix of blue and yellow dots making it look green? Also, the Microsoft campus near where I live is red (Mfg, trade). Wouldn't they be blue (professional)? How about a color for agricultural jobs.

re. microsoft: they are, just bundled into a small area. Look closely.

I really don't understand the classification. Most of Manhattan appears to be predominantly green. And while there's plenty of government/education/healthcare workers here, I don't think that's a reasonable classification for most of the companies with big offices in NYC. Something is off.

Other than it seems to only allows zooming in cities, its data is pretty wrong looking at the local map. I don't see the local manufacturing plant or any farm jobs.

Farm jobs are generally not included in the employment statistics gathered (nor for military occupations) for historical reasons. There was a push in the 80s the change that.

Uhm, agriculture is listed as "Number of jobs in NAICS sector 11 (Agriculture, Forestry, Fishing and Hunting)"[1] which is coded CNS01 in the csv tables[2].

1) http://lehd.ces.census.gov/data/lodes/LODES7/LODESTechDoc7.0...

2) http://lehd.ces.census.gov/data/#lodes

Those still aren't farm jobs. From the LODES description of what it doesn't count:

"Business owners, self-employed persons, some temporary workers, family farmers, military personnel, and others are typically outside of this frame, are not covered, and thus not counted in LODES."

http://www.datafinder.org/metadata/CensusWorkplaceAreaCharac...

For some reason, the map seems to be categorizing technology jobs as green (Healthcare, Education, and Government). Look at SF, and it's almost completely green.

Not necessarily. Here is a list of the largest employers in the Bay area: http://sfced.org/wp-content/uploads/2013/02/LargestEmployers.... The top 5 are healthcare, education, or government.

Given that most of Silicon Valley's employment is not in SF, it's quite likely that the tech jobs in the city are overshadowed by the massive number of education, government, and healthcare jobs.

That's also why Manhattan is mostly green.

The data is very dubious - they've mapped the entire Googleplex as "Retail, hospitality, and other services". The North Bayshore area east of Shoreline is listed as "Manufacturing and trade" - most of that area is Google now.

Sunnyvale and the Stanford area look pretty accurate though, and it got the big office park near Middlefield-Ellis-Whisman and some of the industrial areas in Santa Clara. Perhaps it's working off an old data set (~2000ish)?

Even if it's that old, that doesn't explain why Infinite Loop would be categorized as Manufacturing and trade, with all the area around it as healthcare and government.

Based on industry. It is based on industry not category of job. E.e., everybody at McDonalds is classified as food services even if they are sys admins.

I've heard this as an explanation for the US's perceived lack of "manufacturing" jobs -- if a factory employs its own maintenance people / janitors / whatever, those are manufacturing jobs, but if the factory contracts with another company to send people to do the same work, those are service jobs.

That's kind of odd, that Mass. isn't included. I wonder what happened there.

The Census Bureau is rather vague about that:

> All 50 states, the District of Columbia, Puerto Rico, and the U.S. Virgin Islands have joined the LED Partnership, although the LEHD program is not yet producing public-use statistics for Massachusetts, Puerto Rico, or the U.S. Virgin Islands.

http://lehd.ces.census.gov/

> (Massachusetts has not submitted data to the Census Bureau as of this date)

Presumably not some legal restriction, then. They simply haven't provided the data.

http://lehd.ces.census.gov/research/

Looks like the map is broken right now. Is anyone else having problems viewing it?

Can't see it either. The CSS & JS it loads from Amazon's CloudFront isn't resolving properly.

Broken for me as well, but not cloud front; instead I get this in the console:

    ReferenceError: L is not defined jobs.html:38:12

Noticing an issue with zoom levels. It appears there is only a national zoom resolution, another at the 4-5 states level and then one more resolution at the tiny, neighborhood level. There's not a good resolution for looking at the state level or metropolitan level unfortunately.

Seems to me like modeling every single job based on census data seems unnecessary. Couldn't you also use maybe even just other census data or zoning data on the types of businesses and then pattern based on density, esp. with such rough grouping of classifications?

Great work thought.

Green: Healthcare, Education, and Government

Blue: Professional Services

Looks like how Minneapolis and St. Paul folk perceive each-other is dead on:

http://i.imgur.com/56EgrkE.jpg

How do they perceive each other?

I'm surprised by the prevalence of the green dots, particularly in San Jose.

I'll have to take a look at the data to verify government and education jobs are really that common.

Reply on news.ycombinator.com