Impute

To impute missing data in Vega-Lite, you can either use theimpute transform, either via anencoding field definition or via antransform array.

The impute transform groups data and determines missing values of thekey field within each group. For each missing value in each group, the impute transform will produce a new tuple with theimputed field generated based on a specified imputationmethod (by using a constantvalue or by calculating statistics such as mean within each group).

Documentation Overview

• Impute in Encoding Field Definition
  • Specifying the Key Values to be Imputed
• Impute Transform

Impute in Encoding Field Definition

// A Single View or a Layer Specification{  ...,  "mark/layer": ...,  "encoding": {    "x": {      "field": ...,      "type": "quantitative",      "impute": {...},               // Impute      ...    },    "y": ...,    ...  },  ...}

Anencoding field definition can include animpute definition object to generate new data objects in place of the missing data.

Theimpute definition object can contain the following properties:

Forimpute in encoding, the grouping fields and the key field (for identifying missing values) are automatically determined. Values are automatically grouped by the specified fields ofmark property channels,key channel anddetail channel. If x-field isimputed, y-field is the key field. Basically, any missing y-value in each group will lead to a new tuple imputed, and vice versa.

In this example, weimpute they-field ("b"), so thex-field ("a") will be used as the"key". The values are then grouped by the field"c" of thecolor encoding. The impute transform then determines missing key values within each group. In this case, the data tuple where"a" is3 and"c" is1 is missing, so a new tuple with"a" =3,"c" =1, and"b" =0 will be added.

Besides imputing with a constantvalue, we can also use amethod (such as"mean") on existing data points to generate the missing data.

Theframe property ofimpute can be used to control the window over which the specifiedmethod is applied. Here, theframe is[-2, 2] which indicates that the window for calculating mean includes two objects preceding and two objects following the current object.

Specifying the Key Values to be Imputed

Thekeyvals property provides additional key values that should be considered for imputation. If not provided, all of the values will be derived from all unique values of thekey field. If there is no grouping field (e.g., nocolor in the examples given above), thenkeyvalsmust be specified. Otherwise, the impute transform will have no effect on the data.

Thekeyvals property can be an array:

Alternatively, thekeyvals property can be anobject defining a sequence, which can contain the following properties:

Impute Transform

An impute transform can also be specified as a part of thetransform array.

// A View Specification{  ...  "transform": [    ...    {      // Impute Transform      "impute": ...,      "key": ...,      "keyvals": ...,      "groupby": [...],      "frame": [...],      "method": ...,      "value": ...    }    ...  ],  ...}

For example, the same chart withimpute in encodingabove can be created using theimpute transform. Here, we have to manually specify thekey andgroupby fields, which were inferred automatically forimpute inencoding.

Similarlykeyvalsmust be specified if thegroupby property is not specified.