This section explains how to create and configure your data cubes.
Create a new data cube from the home view by clicking the + Data cube
button.
The Create new data cube
modal allows you to define the source data for a data cube.
Checking the Auto fill dimensions and measures
checkbox will automatically detect your schema
and create dimensions and measures accordingly.
Another way to create a new data cube is to duplicate an existing data cube and edit its properties.
After creating the data cube you will be taken to the edit screen to adjust any details.
Click the pencil icon in the data cube header to edit a data cube.
Within the edit view you can change the title, description, and default timezone of the data cube. You can also edit and create dimensions and measures from their respective tabs.
Pivot will automatically detect the columns in your data source and recommend some possible dimensions and measures.
Schema detection can be helpful when you are creating a new data cube in which case it will allow you to initialize the data cube with some dimensions and measures.
The other use for schema detection is when you evolve your schema and add columns over time.
If you have just added a column and you want a corresponding dimension to be created for it, you can use schema detection to automatically construct that column for you.
This functionality is accessible from the Suggestions
tab of the New dimension and New measure modals.
Pivot looks at the dataset metadata and uses the returned list of columns, their types, and their aggregation (in case of rollup) to determine what dimensions and measures to suggest.
Pivot generates dimensions and measures by applying the following rules to the discovered underlying column types:
SUM
measure or an otherwise appropriate measure if the column is marked as being aggregated as part of rollup.countDistinct
measure.While schema detection is an invaluable tool for quickly getting you up and running with a new data source, it can never do a perfect job.
Once you create a data cube you should play with it and tailor it to be perfectly suited to your needs.
Don't hesitate to change and delete the auto-generated dimensions and measures - they will always be there in the Suggestions
tab.
In particular, schema detection can not detect these common scenarios:
countDistinct
- number of unique values.The following advanced options are available when you expand the More
chevron.
Subset filter formula: Specify a plywood formula that applies a mandatory, hidden filter to all queries made through this data cube. See row level restriction.
Required filter token: Specify a token that corresponds to a filter token defined in a user role. See token based row level access control.
Primary time dimension: Specify the time dimension that corresponds to when the data point was actually added. This is used for checking the freshness of data and also potentially enforcing a time filter. Defaults to Time
.
Enforce time filter: If set to true, ensures that every query is filtered on the primary time dimension. This should be used if time unbounded queries are likely to be slow due to the volume and tearing of data.
Instances: Configure multiple filtered instances within a single data cube via a configuration object. See example configuration.
Refresh time: Configure how the data cube will calculate the data's latest time.
Defaults to Query
, which means it will query the data to learn about the latest data ingested. This can also be set to predefined in which case latest current time will always be assumed. An alternative 'fixed' time can also be defined as the 'fixedMaxTime' property in the options.
Query caching: Specify query caching behavior. Allowing caching can greatly speed up exploration but can also cause results to be a little out of date especially in realtime rolled up datasets.
Additional data cube configuration options as JSON with any of the following properties:
fixedMaxTime
: Specify a fixed time value for data cubes to use as the latest "max time" instead of querying for it or using the current time. This configuration should be used in conjunction with the Refresh time
setting set to 'fixed'.timeoutOverride
: Override a connection-wide timeout for queries made through this data cube.exactResultsOnly
: If true, disallows approximate aggregators and force Pivot to make groupBy over topN queries.customAggregations
: Map any Druid aggregation to a Pivot measure. Aggregation definitions should be keyed on measure names.customTransforms
: Map any Druid extraction function function to a Pivot dimension. Extraction function definitions should be keyed on dimension names.crossTabBatchSize
: Configure the number of cross tab rows loaded at once (defaults to 80
)piiMask
: Define a threshold below which data is considered Personally identifiable information (PII) and hidden. See example.Each section of presets in the time filter menu can be configured with custom presets. See an example of latest presets.
latestPresets
: An array of preset objects that will override the LATEST DATA
presets in the time filter menu.currentPresets
: An array of preset objects that will override the CURRENT
presets in the time filter menu.previousPresets
: An array of preset objects that will override the PREVIOUS
presets in the time filter menu.splitLimits
: Provide an array of numbers that will override the default options of [5, 10, 25, 50, 100]
shown in the limit dropdown of the split menu.ignoreDimensionInExplain
: Provide a list of dimensions to ignore when computing an explanation.rankExpression
: Configure the measure whose value is used to order dimension values in the dimension filter menu. Defaults to count
.druidContext
: An object that will be passed in as the Druid query context.priority
: A number that will be passed as 'priority' into the Druid query context.maxDownloadLimit
: Configure a limit (in number of rows) to downloads.largeDownloadInterval
: Configure a limit (as a time interval) to downloads.suppressOverallByDefault
: If true, hides the "overall" element of a visualization by default.boostPrefixRank
: If true, will "boost" dimension values where matches occur at the beginning of the string over values with matches found in other positions in the dimension filter menu.boostSelectedMeasures
: If true and the data cube view is in multi-measure selection mode, will "boost" selected measures to the top of the measures list in the measure picker menu.grayOutTrivial
: If true, renders "trivial" dimensions in gray in the dimensions panel.
A continuous dimension is trivial if it has a max value equal to its min value.
A discrete dimension is trivial if it has less than 2 values.alwaysShowCompareControl
: If true, will remove the ability to toggle time compare.disablePreview
: If true, disables the preview of values in the dimension and measure modals.showExplainBadge
: If true, shows a badge with the standard deviation for the sample values used for the explanation.{
"dimensionFormula": "$language",
"values": [
{
"title": "English",
"value": "en"
},
{
"title": "Chinese",
"value": "zh"
},
{
"title": "Spanish",
"value": "es"
}
]
}
dimensionFormula
is a plywood formula that operates on a dimension.
values
is a list of objects with keys of title
, value
where value
is the value that will be matched with the dimensionFormula
expression and title
is the label displayed in the UI.
{
"piiMask": {
"aggregate": "$main.count()",
"threshold": 10,
"maskText": "HIDDEN [%i]",
"otherText": "Other (%i)",
"hideAggregates": false
}
}
Note that %i
will be resolved as the row index of the datum
{
"latestPresets": [
{ "name": "1M", "selection": "$m.timeRange('PT1M', -1)" },
{ "name": "5M", "selection": "$m.timeRange('PT1M', -5)" },
{ "name": "30M", "selection": "$m.timeRange('PT1M', -30)" },
{ "name": "1H", "selection": "$m.timeRange('PT1H', -1)" },
{ "name": "6H", "selection": "$m.timeRange('PT1H', -6)" }
]
}
name
corresponds to the UI label of the preset.
selection
corresponds to the interval of time selected.
Note that in the selection
value, $m
will be resolved as the data's max time and $n
as the current time.