One of the UTAs told me that there is a misspelling of a key in the
County Demographics data set. It is:

"Income"-->"Median Houseold Income"

This is a feature request. Can you add the get method that's available in the BlockPy version of CORGIS to the Python version?

If I use CORGIS in BlockPy, there's this handy .get method that takes in an attribute, filter column, and filter value, and returns a list of data values with the correct type.

If I use the Python version, I only have the get_XXX method, e.g. get_report in the AIDS dataset. This returns a complex Python object, but it doesn't have the .get method (unless I'm somehow missing it - it just seems to be a dictionary).

Why this matters: My students often start their projects in BlockPy, but some of them realize they want functionality (e.g. writing a file) outside of BlockPy's abilities, so they download their script. When they do, it suddenly doesn't work anymore. Even if they download the CORGIS .data and .py files and put them in the same directory, the API is different. And the .get method is a nice utility function even for advanced programmers, so why not include it in each .py file?

Thanks!

https://rpubs.com/neros/61800

Medal of Honor: data is fine, it might be good to note in the description that a value of -1 means a missing value, since people who have never coded before may not know that is somewhat standard.

Music: Data is not fine. There is a field under artist called “hotttnesss” which is a kind of weird way to spell it. It seems like in this dataset, unknown int values default to 0, which should probably be noted in the description (especially if there is no universal int value for unknown). The description under the dropdown menu for Artist.similar is “unknown” and a message pops up saying “Given this filter, there was no variation in the data - visualization will be empty”. Artist.terms_freq is all 0 values. Release.name’s description is “unknown”, and I can’t seem to make a graph that makes proper use of the field. Song.artist _mbtags’ description is “unknown”, all values are 0. For song.hotttnesss (again, a weird way to spell it), there are a lot of -1 values and 0 values- does this mean both ints were used as unknown, or is there something else I don’t understand about that? Song.mode values are all 0. Song.title values are all 0, which makes sense because it would be strings, but I can’t find a way to use the song.title field because the only bar chart “group by” fields are artist.name and song.year. For song.year it should be noted somewhere that unknown values are 0.

Opioids: data is all ok. Might be a good idea to note that the x axis for line plots is years since 1999.

Police shootings: dataset is pretty sparse but the data is ok.

These aren't working anymore since (I think) today:

https://think.cs.vt.edu/corgis/csv
https://think.cs.vt.edu/corgis/json
https://think.cs.vt.edu/corgis/java

https://github.com/washingtonpost/data-school-shootings

This seems like a rough option for Virginia Tech. I mean, we do have some pretty dark datasets, but the local context is sensitive here. But is that a reason not to address something in a college setting? Need to have more conversation about this.

Does this mean it installs all datasets?

Or perhaps installing Corgis allows you to launch an interactive installer util?

Or perhaps we can send parameters to Pip?

Currently, unknown genders are empty strings. We should make them "unknown" or something more descriptive.

I've been looking at the old repo trying to figure out how to use it and the 'contributor.md' was helpful and the issues list lead me here. I'd like to become a contributor and start by helping with the csv file that needs commas I found.

My main question is how do I as a new user access the billionaires list or run it as a data set for just that one database?

How often is that database updated or can I access just one?

Would like to contribute where I can.

All of the values in the Engine Information.Hybrid column are "True", even though some (many/most?) of the cars are not hybrids.

Somewhere over the years, we lost the docs on the Disease dataset. It's now called the Health dataset. Here's my notes on it from before:

https://www.tycho.pitt.edu/
This library provides data about reported ocurrences of diseases in American history over time. Some diseases (e.g., Measles) have more information and stretch back farther than others (e.g., "AIDS"). Some places seem to have missing information (either that, or the population of Virginia was immune to Measles until the 1920s).

Cancer Dataset explorer has a problem: report["Rates"]["Age and Sex"]["45 - 64"] (or any other 3 categories) do not display example value (it is a dict with two keys: “Female” and “Male”, each hold a float value)

I also think this dataset might just be hot garbage and needs to be reworked from the ground up. I see this was an issue dating back to the old version.

Both the documentation and the .csv file lack mentions of portion size, which makes it impossible to interpret nutritional content figures.
(e.g. raw blueberries are stated to contain 9.7mg of Vitamin C, but how much fruit are we talking about exactly?)

A cursory examination of the documentation accompanying the USDA FNDDS database suggests the raw figures are based on 100g portions. It appears that the csv file uses the same (but I don't know for sure).

This isn't a network issue, it's client side. Very big datasets can take several seconds to render. Since this is happening on page load, it feels like the system is lagging. We should do a repaint before actually loading the dataset, have a loading symbol during dataset switches, and probably do find some way to let it happen in the background more (suspensions? promises? web workers?).

User report:

I wanted to let you know that the descriptions for this data set for the *.Value and *.Quality are incorrect. Specifically, *Quality should be Worse, Average, Better and *.Value seems to be Lower, Average, Higher.
https://corgis-edu.github.io/corgis/csv/hospitals/