nychealth/EH-dataportal

Data explorer enhancements - ideas

mmontesanonyc opened this issue · 2 comments

mmontesanonyc commented

We'll add a list of data explorer enhancements in this issue. When one gets traction, we'll spin off a new, focused and dedicated issue for it.

Console logs: Revise console logs to be descriptive. Log the relevant objects, specific selections. Partly, this can be useful for us (easier to find the names of objects), and partly, useful to any techie poking around.

Map:

  • Rate / Count toggler, with count view showing a bubble map (see #997)
  • #1003
  • #616
  • Allow turning off geo types (in indicators.json)
  • #97 .
  • #497 - Min/max on maps? See Restaurants with A Grades - think about showing change over time via maps.
  • Manually override automatic breaks for color scale
  • Vertical line on bar chart to better highlight mouseover neighborhood (esp. when value = 0)

Table:

  • "Select All" button for years
  • Refine styles: smaller text; better headers, colors, and ledger lines; render a column as a bar charT?

Links

  • #106
  • Button-out the drop-down menu? Or some other way that people can see all the potentially connected things?
  • Distinguish overlapping points

Trend

  • #840
  • Click on group point or legend highlights that group (like links chart)
  • Show missing years as explicitly missing. Using end_period for x-axis field, but TimePeriod for label won't work, because labelExpr doesn't have access to anything but the name and value of the field. One alternative: finding missing sequence values in renderComparisonsChart and adding them to the data. Another alternative: using end_period, but conditionally modifying the timeUnit or format in renderComparisonsChart to recreate TimePeriod. This is probably the better option.

Disparities

  • ICE score
  • TRIE neighborhoods,

General

  • Consider adding a "threshold" data type to support indicators where we have a single value as a measure (like a regulatory threshold or some other reference value). This could perhaps be in vizOptions in indicators.json, and could be paired with importing it into the page in a variety of places (as a y-value in the trend chart; as a stick-to-top value in the table; in the tooltip on the map?) to add explanation/comparison/context in a strong way.
  • What if, when things are inactive (e.g., no map, no trend), instead of just having it inactive - we explain our rationale? Like, "We don't map things if we don't have data finer than RESOLUTION."
  • See asthma - use "label" for indicator menu if Indicator header exists?
  • Database rebuild - add facets: age. sex. other?
  • Data updated time, using git log on EHDP-data repo

Data Explorer Drill-down
Add some type of interpretation - something that helps answer The Why.

  • Perhaps a "what does this chart mean"
  • Something customized and specced at the indicator level where we link to the top resources on the portal instead of relying on the Related footer
  • A threshold value; an interpretative sentence for a chosen neighborhood that compares the value to the NYC average.
mmontesanonyc commented

NB, consolidated with other "Ideas" card. FYI, Mike had some thoughts (added here) about adding some type of "Data Explorer Drill-down" functionality that gets into ways to explain what a user is seeing. Touches on a lot of ideas we had - and definitely an interesting area for growth.

mmontesanonyc commented

@cgettings related to WM's point about legend labels not wrapping. I've been interested for a while in applying direct labels to trend charts, instead of legends. Here's an initial - there's a layer that takes the max date, the value at the max date, and then encodes a text mark there. The major drawback is interference, but I'm going to look into the label transform and avoidMarks to see what can be accomplished with that.