Closed gcgruen closed 8 years ago
Not having been at the lecture with Richard, hexagonal graphs are kind of tough for the layperson to understand at an entry level -- I think it depends on who your target audience is for these vizes. Would it be possible to somehow narrow your dataset further, like for whom the data is bad, or somehow aggregate by income level or school size or region or something? Maybe make your dots smaller?
Does the dataset identify schools, or is it anonymous? (It might be interesting to investigate some of the outliers if so. Maybe for another project.)
It's a great idea to show black students support in education equality. Since the scatter points overlap a lot, it might be better to make bar charts which show the total number of enrolled white/black students. If you would like to describe the enrolment ratio, you may want to calculate first, then make a line graph about the white/black ratio variation in the past decade(whatever period you want). I'm also very curious about the outliers.
Thank you @shuyao1201 :)
Update: First graphic
I love how the visuals have taken a really different turn as we have understood this dataset better! Hurrah for figuring out what works. Overall I think it's great, and it really clearly communicates the patterns you found.
I dislike the bolded words in the annotations, I think it's distracting and doesn't add to the clarity of the explanation. The ticked lines seem a little odd. Are they separated, or -*- or somesuch thing? They just look weird to me :/
Thanks for your feedback! I changed most of the boldness and also the stroked line weight. That really made it better, so thanks for pointing this out!
Another form of graphical representation of the same data -- slight difference: includes the absolute counts of schools. Disadvantage: less statistical metrics. Go with both?
Updated existing graphics with corrected description and adding further graphics on advanced coursework
(more data on police in schools, and how the decision is made to place an officer in a school)
From the Department of Justice:
School Safety Data Sources There are a number of national sources of school safety data. Data are often broken into categories, such as urban/rural; age groups; male/female. These can be helpful in identifying where a school stands compared to other schools with similar characteristics. National data sources: • National Crime Victimization Survey (NCVS) • School Crime Supplement (SCS) to the National Crime Victimization Survey • School Survey on Crime and Safety (SSOCS) • Schools and Staffing Survey (SASS) • Youth Risk Behavior Surveillance System (YRBSS) For state data, the state attorney general or child services agency can often provide information. Locally, school districts, law enforcement, social service agencies, and colleges and universities can be useful sources.
Thanks for the input. I think I'll leave it to a future project to combine this dataset with another dataset ;) Also, as I understand from the documentation, the next release with the data of 2015/2016 will include the actual count of law enforcement officers plus the count for security guards.
Adding last graphic on law enforcement officers in relation to racial breakdown of student population.
Closing since pull request #265 has been accepted
Scope:
Collaborating with @kromreig and @HarshaDevulapalli , I would like to provide visuals for some of the claims made by the Black Lives Matter movement:
"Black students are
(Ugly) Proof of viability:
To see whether I can get sth out of the data, I focused on 1) the enrolment in gifted and talented programs Male Students:
![gt_wm](https://cloud.githubusercontent.com/assets/7279358/17720849/3cd21e7a-63f3-11e6-8251-0c433e9e2929.png)
Female Students:
![gt_w_f](https://cloud.githubusercontent.com/assets/7279358/17720852/44ef8098-63f3-11e6-80d2-bd2bc787caaa.png)
Data:
The dataset is huge (more than 95,000 rows), so only working with a relevant subset of columns The overall dataset can be found here
Headlines:
Questions:
That's why I was looking around for alternatives. In the description for hexagonal bin plots it says that it is convenient for very dense datapoints -- so this case, I guess? How do people feel about this kind of graph? Maybe not with a sequential scale (as the example in the documentation, but a diverging one)
Story checklist:
My pitch issue was: #219
PITCHING.md
)