adilyalcin / keshif Goto Github PK
View Code? Open in Web Editor NEWKeshif - Data Made Explorable (Prototype)
Home Page: https://keshif.me
License: BSD 3-Clause "New" or "Revised" License
Keshif - Data Made Explorable (Prototype)
Home Page: https://keshif.me
License: BSD 3-Clause "New" or "Revised" License
Making the filtered time text visible always will help user see state.
Making it editable directly will make it easy to set exact time ranges.
Currently it's not so well designed...
Filter search box: when user pressed escape, the action propagates to top and all filters are cleared.
Make sure you only clear the text box, not the whole chart filter selection.
When clicked, the chart should take all screen, re-adjust itself, and make the background transparent.
For examples, check http://datawrapper.de/gallery/
Once filtering is applied, the grouping of items are broken (the line seperation based on matching sort values)
List display should only display about 100 items at a time, and allow forward-backford,etc in the list
The list control should be added to bottom of the display, and look like
|____________________________________________________|
| First | Prev | | Next | Last |
And include a dragbar too in the middle to make it more "visual" ? <-----------X-------->
I have selectiong box and i want to append timeItemMap attribute on different category.
Please help.
You can currently only drag the handles, you can do so by clicking on the range bars too.
Example: Bederson pub page - step size doesn't always really work well, jumps 2 years instead of 1 sometimes, and sometimes no year… hmmm...
Scroll handlers can just stick too much and become annoying.
Put a better summary.
Put better description of possible actions.
Smaller credits.
Summary blocks may be re-ordered by dragging them.
Design idea:
When mouse is over a categorical widget, show a E-like reorder (click and drag) button on the right side of the header.
The arrows will not be visible by default.
Also align the button on the left-hand side of the facet always.
Currently, a page can only include a single keshif instance
This is important if the loader fails, you still provide access to the data.
Currently, there must be at least one sorting option.
What does it mean to specify no explicit sorting option for the list? Should it be, by design, not allowed?
Can user specify the sorting function, but not display a sorting column?
Example dataset: Transactions In Major League Baseball
Tables are loaded twice, since filter is working in two ways (from/to team)
Currently, there is no way to focus on a specific date range ( year, etc) by directly clicking on the range itself. You need to DRAG min & maximum, which is actually probably not the most effective use case.
With this change, support clicking on a time label (year, month, day, etc), and select the time range from that label to the next label in display.
The labels may not include every year, so selecting only 1-unit time is not really possible, depending on the scale of time axis. Selecting the range-between-labels & then dragging and adjusting the selection is probably going to work better.
Comments, questions? Drop them below. Thanks,
Currently, it's only zero or one. The only thing to work on is to compute min/max given multiple time attrib per item (for setting timechart boundaries)
The motivation is that some filters are very connected (number of yea's and number of nay's).
Currently, each facet bar chart is scaled independent of other facets, thus creating different scales per each facet.
Design: Only chars that are above/blow each other should be the same.
A lot of people click on year instead of dragging the range. May support like drag&specify
Currently the interface just stops.
We can add +/- buttons when appropriate so that the user can see more categories.
Question is: is it necessary, how to implement and design this.
How does manual adjustment interact with re-layout algorithm?
Currently, if multi selection is supported (by default), selection works as an or. Allow selection of "and" when multiple selections are made.
Not using external img files, but by referencing by dynamically generated DOM elements...
The bar behind dots can make readability harder (when dots are many & transparent setting is used).
Also, it may not be necessary in some cases, where first/last item is no really important.
Currently it doesn't work.
I think the ipad/touchscreen scrolling issues are also related to this.
SOLR is an Apache Lucene Based search engine supporting facets http://lucene.apache.org/solr/
Internet Explorer (10) does not seem to deal with html blocks inside svg well. Actually, other browsers cannot apply transformations too. But IE just doesn't display it.
Result: Headers in facets are not shown.
Code needs to be restructured so that html part is outside the svg definition per facet...
Currently, single select vs multiselect charts show no visual difference in interaction for the user.
If users can notice the difference clearly (if it doesn't feel intuitive), Users may feel a little out-of-control (need to test?). Some for of visual distinction between two modes can also make it more confusing to people who don't know the difference.
Currently it is just a small bar. It may not be really visible. What's the best way of showing this?
Maybe, make the bar chart bigger, and let it expand over sort column. "Number of items" will move right
Probably by clicking on title, you can present options to user to select which filterthey want to use. Thus, allow, changing the filters seen on screen.
Applies to time as well. Only time-relevant attributes whould be shown in the dropbox.
Currently, you cannot ask the viz to show "items without any value under a filter". It's similar to asking "items with zero attributes", and you can basically ask "item with one attribute - any matching (from multiple or selection)".
Chart config option: Enable/disable "none" category mapping (off by default).
Currently, items are sorted linearly, one item per row.
Another common approach is having a matrix grid, and have multiple items per row.
How would that compare?
-- findthebest.com also offers two options: list and grid view
-- Amazon item lists and many other shopping / listing sites use matres as default (more items on screen given a wide page width)
All areas in chart areas should be un-selectable (except list display) It can be annoying for users.
Currently, all the categories are correctly highlighted on mouse over.
Only time attribute is not highlighted
Need to register touch events for scrolling and selection.
If chart height is the same, skip that chart, but maintain a "dirty" flag so that whenever chart is visible again, it gets re-computed correctly.
Seems simple.
Requires state serialization & initialization.
There are parts which directly assume all data is "selected" on page load.
This may work better in datasets where there are few items under each category, and you want to get a scrolling overview and reduce repetition of group name for each item displayed.
So, the change suggests support for the following
| ** GroupName ----------------------------------------------------|
| Item Info -----------------------------------------------------------|
| Item Info -----------------------------------------------------------|
| Item Info -----------------------------------------------------------|
instead of
| GroupName | Item Info ------------------------------------------|
| GroupName | Item Info ------------------------------------------|
| GroupName | Item Info ------------------------------------------|
When all item counts are zero, no bar is displayed.
When range is small (0-1 for example), the labels start showing 0.2, 0.4, 0.6, 0.8 ...
But, such valuea may be allowed if the value shown is an aggregate of items that are mapped under the facet (total downloads in a specific category, for example).
Current implementation doesn't allow this. It's a common practice.
Maybe there can be some common-sense automation?
The animation should yet be applied when height layout is changed, which may affect the scrollbar position within chart.
Inactive ones are the ones you cannot click / will make zero result sets. They are hidden/removed automatically (in most cases).
UI can include
Add an API option to turn this on/off s facet config. UI part will dynamically change this setting.
Google updated their spreadsheet interface.
Spreadsheets created in their new format cannot be accessed by google vis API tools used by Keshif.
Once google updates their documentation regarding new access paths for new documents, I will update Keshif / add an option to support new API.
A declarative, efficient, and flexible JavaScript library for building user interfaces.
🖖 Vue.js is a progressive, incrementally-adoptable JavaScript framework for building UI on the web.
TypeScript is a superset of JavaScript that compiles to clean JavaScript output.
An Open Source Machine Learning Framework for Everyone
The Web framework for perfectionists with deadlines.
A PHP framework for web artisans
Bring data to life with SVG, Canvas and HTML. 📊📈🎉
JavaScript (JS) is a lightweight interpreted programming language with first-class functions.
Some thing interesting about web. New door for the world.
A server is a program made to process requests and deliver data to clients.
Machine learning is a way of modeling and interpreting data that allows a piece of software to respond intelligently.
Some thing interesting about visualization, use data art
Some thing interesting about game, make everyone happy.
We are working to build community through open source technology. NB: members must have two-factor auth.
Open source projects and samples from Microsoft.
Google ❤️ Open Source for everyone.
Alibaba Open Source for everyone
Data-Driven Documents codes.
China tencent open source team.