birm / datadash

Data Exploration Dashboard Platform

DataDash

DataDash is a specification for interactive dashboards. This repository includes an engine which serves to the specification, and a display which consumes the specification.

DataDash API Version 1

Filter

Many queries require a filter. A filter is an object where the keys are the names of columns, and the values are objects describing the rules to filter that column on. V1 supports four filter operations in each column:

• match: true if the value literally matches
• regex: true if the regex returns true
• less: true if the value is less
• greater: true if the value is greater.

These can be stacked, such as height:{less:5,greater:1} which returns all values between 1 and 5, noninclusive.

Records meet the filter if ALL filter rules are met.

Methods

The following methods are part of the V1 Specification. Implementations may have additional parameters or methods, for authenticaiton, for example.

/v1/datasets

Returns a list of datasets avaliable

Engines can support more than one dataset. As a result, calls to any data-fetching method should specify which dataset it uses. This call allows a user to see the options avaliable.

Params

None

Returns

List of strings, each being a dataset.

/v1/summary

Returns a list of columns (keys) in the dataset

Some operations require specifying a column; this lets users see which columns are avaliable for a dataset.

Params

• dataset -- the selected dataset

Returns

List of strings, each being a column in the dataset.

/v1/tabular

Returns data in tabular form matching the filter

Some operations require specifying a column; this lets users see which columns are avaliable for a dataset.

Params

• dataset -- the selected dataset
• start -- the number of record to start on (beware numbers will change with change in filter)
• len -- the number of records to return
• filter -- The active filter

Returns

List of objects, each being a record.

/v1/counteach

Returns how many records match the filter for each value of a column

Params

• dataset -- the selected dataset
• col -- the column to get the count on
• filter -- The active filter

Returns

An object: keys are values of the specified column, values are counts matching the filter.

datadash's Issues

Basic Interactive Frontend

Sampling Support

Maybe support deterministic random sampling given seed.
Something requested is to find a subset of a population matching a different population.
I’m not sure how I want to go about this, but some ideas:
Knn without replacement
a is set to match, b is set of candidates to use, b’ is selected set of b.
Start with b’ empty
For each in population a, find nearest in population b, add to b’, remove from candidate pool, repeat
May want to set a threshold over which to report seperately

Document APIs for Standard

Filters

1d

• QQ against normal
• hist
• bar

2d/3d

• qq between variables
• time series
• scatter

Visualizations

• Table
• Image Grid
• SPLOM

Mongo DD Engine