ROLLINGAVERAGE Function

Computes the rolling average of values forward or backward of the current row within the specified column.

• If an input value is missing or null, it is not factored in the computation. For example, for the first row in the dataset, the rolling average of previous values is the value in the first row.

• The row from which to extract a value is determined by the order in which the rows are organized based on the order parameter.

• If you are working on a randomly generated sample of your dataset, the values that you see for this function might not correspond to the values that are generated on the full dataset during job execution.

• The function takes a column name and two optional integer parameters that determine the window backward and forward of the current row.

• The default integer parameter values are -1 and 0, which computes the rolling average from the current row back to the first row of the dataset.

• This function works with the Window transform. See Window Transform.

For more information on a non-rolling version of this function, see AVERAGE Function.

Wrangle vs. SQL: This function is part of Wrangle, a proprietary data transformation language. Wrangle is not SQL. For more information, see Wrangle Language.

Basic Usage

Column example:

rollingaverage(myCol)

Output: Returns the rolling average of all values in the myCol column.

Rows before example:

rollingaverage(myNumber, 3)

Output: Returns the rolling average of the current row and the three previous row values in the myNumber column.

Rows before and after example:

rollingaverage(myNumber, 3, 2)

Output: Returns the rolling average of the three previous row values, the current row value, and the two rows after the current one in the myNumber column.

Syntax and Arguments

rollingaverage(col_ref, rowsBefore_integer, rowsAfter_integer) order: order_col [group: group_col]

Argument

Required?

Data Type

Description

col_ref

Y

string

Name of column whose values are applied to the function

rowsBefore_integer

N

integer

Number of rows before the current one to include in the computation

rowsAfter_integer

N

integer

Number of rows after the current one to include in the computation

For more information on the order and group parameters, see Window Transform.

col_ref

Name of the column whose values are used to compute the rolling average.

• Multiple columns and wildcards are not supported.

Usage Notes:

Required?

Data Type

Example Value

Yes

String (column reference to Integer or Decimal values)

myColumn

rowsBefore_integer, rowsAfter_integer

Integers representing the number of rows before or after the current one from which to compute the rolling average, including the current row. For example, if the first value is 5, the current row and the five rows before it are used in the computation. Negative values for rowsAfter_integer compute the rolling function from rows preceding the current one.

• rowBefore=0 generates the current row value only.

• rowBefore=-1 uses all rows preceding the current one.

• If rowsAfter is not specified, then the value 0 is applied.

• If a group parameter is applied, then these parameter values should be no more than the maximum number of rows in the groups.

Usage Notes:

Required?

Data Type

Example Value

No

Integer

4

Examples

Tip

Example - Compute prior quarter values

This example covers how to use the NEXT function to create windows of data from the current row and subsequent (next) rows in the dataset. You can then apply rolling computations across these windows of data.

Functions:

Item

Description

NEXT Function

Extracts the value from a column that is a specified number of rows after the current value.

ROLLINGAVERAGE Function

Computes the rolling average of values forward or backward of the current row within the specified column.

NUMFORMAT Function

Formats a numeric set of values according to the specified number formatting. Source values can be a literal numeric value, a function returning a numeric value, or reference to a column containing an Integer or Decimal values.

Source:

The following dataset contains order information for the preceding 12 months. You want to compare the current month's average against the preceding quarter.

Date

Amount

12/31/15

118

11/30/15

6

10/31/15

443

9/30/15

785

8/31/15

77

7/31/15

606

6/30/15

421

5/31/15

763

4/30/15

305

3/31/15

824

2/28/15

135

1/31/15

523

Transformation:

Using the ROLLINGAVERAGE function, you can generate a column containing the rolling average of the current month and the two previous months:

Transformation Name Window ROLLINGAVERAGE(Amount, 3, 0) -Date

Note the sign of the second parameter and the order parameter. The sort is in the reverse order of the Date parameter, which preserves the current sort order. As a result, the second parameter, which identifies the number of rows to use in the calculation, must be positive to capture the previous months.

Technically, this computation does not capture the prior quarter, since it includes the current quarter as part of the computation. You can use the following column to capture the rolling average of the preceding month, which then becomes the true rolling average for the prior quarter. The window column refers to the name of the column generated from the previous step:

Transformation Name Window NEXT(window, 1) -Date

Note that the order parameter must be preserved. This new column, window1, contains your prior quarter rolling average:

Transformation Name Rename columns Manual rename window1 'Amount_PriorQtr'

You can reformat this numeric value:

Transformation Name Edit column with formula Amount_PriorQtr NUMFORMAT(Amount_PriorQtr, '###.00')

You can use the following transformation to calculate the net change. This formula computes the change as a percentage of the prior quarter and then formats it as a two-digit percentage.

Transformation Name New formula Single row formula NUMFORMAT(((Amount - Amount_PriorQtr) / Amount_PriorQtr) * 100, '##.##') 'NetChangePct_PriorQtr'

Results:

Note

You might notice that there are computed values for Amount_PriorQtr for February and March. These values do not factor in a full three months because the data is not present. The January value does not exist since there is no data preceding it.

Date

Amount

Amount_PriorQtr

NetChangePct_PriorQtr

12/31/15

118

411.33

-71.31

11/30/15

6

435.00

-98.62

10/31/15

443

489.33

-9.47

9/30/15

785

368.00

113.32

8/31/15

77

596.67

-87.1

7/31/15

606

496.33

22.1

6/30/15

421

630.67

-33.25

5/31/15

763

421.33

81.09

4/30/15

305

494.00

-38.26

3/31/15

824

329.00

150.46

2/28/15

135

523.00

-.74.19

1/31/15

523

Example - Rolling window functions

This example describes how to use rolling computational functions.

Functions:

Item

Description

ROLLINGSUM Function

Computes the rolling sum of values forward or backward of the current row within the specified column.

ROLLINGAVERAGE Function

Computes the rolling average of values forward or backward of the current row within the specified column.

ROWNUMBER Function

Generates a new column containing the row number as sorted by the order parameter and optionally grouped by the group parameter.

Also:

Item

Description

MONTH Function

Derives the month integer value from a Datetime value. Source value can be a a reference to a column containing Datetime values or a literal.

FLOOR Function

Computes the largest integer that is not more than the input value. Input can be an Integer, a Decimal, a column reference, or an expression.

ROWNUMBER Function

Generates a new column containing the row number as sorted by the order parameter and optionally grouped by the group parameter.

The following dataset contains sales data over the final quarter of the year.

Source:

Date

Sales

10/2/16

200

10/9/16

500

10/16/16

350

10/23/16

400

10/30/16

190

11/6/16

550

11/13/16

610

11/20/16

480

11/27/16

660

12/4/16

690

12/11/16

810

12/18/16

950

12/25/16

1020

1/1/17

680

Transformation:

First, you want to maintain the row information as a separate column. Since data is ordered already by the Date column, you can use the following:

Transformation Name Window ROWNUMBER() Date

Rename this column to rowId for week of quarter.

Now, you want to extract month and week information from the Date values. Deriving the month value:

Transformation Name New formula Single row formula MONTH(Date) 'Month'

Deriving the quarter value:

Transformation Name New formula Single row formula (1 + FLOOR(((month-1)/3))) 'QTR'

Deriving the week-of-quarter value:

Transformation Name Window ROWNUMBER() QTR Date

Rename this column WOQ (week of quarter).

Deriving the week-of-month value:

Transformation Name Window ROWNUMBER() Month Date

Rename this column WOM (week of month).

Now, you perform your rolling computations. Compute the running total of sales using the following:

Transformation Name Window ROLLINGSUM(Sales, -1, 0) QTR Date

The -1 parameter is used in the above computation to gather the rolling sum of all rows of data from the current one to the first one. Note that the use of the QTR column for grouping, which moves the value for the 01/01/2017 into its own computational bucket. This may or may not be preferred.

Rename this column QTD (quarter to-date). Now, generate a similar column to compute the rolling average of weekly sales for the quarter:

Transformation Name Window ROUND(ROLLINGAVERAGE(Sales, -1, 0)) QTR Date

Since the ROLLINGAVERAGE function can compute fractional values, it is wrapped in the ROUND function for neatness. Rename this column avgWeekByQuarter.

Results:

When the unnecessary columns are dropped and some reordering is applied, your dataset should look like the following:

Date

WOQ

Sales

QTD

avgWeekByQuarter

10/2/16

1

200

200

200

10/9/16

2

500

700

350

10/16/16

3

350

1050

350

10/23/16

4

400

1450

363

10/30/16

5

190

1640

328

11/6/16

6

550

2190

365

11/13/16

7

610

2800

400

11/20/16

8

480

3280

410

11/27/16

9

660

3940

438

12/4/16

10

690

4630

463

12/11/16

11

810

5440

495

12/18/16

12

950

6390

533

12/25/16

13

1020

7410

570

1/1/17

1

680

680

680

Example - Rolling computations for racing splits

This example describes how to use rolling statistical functions.

Functions:

Item

Description

ROLLINGAVERAGE Function

Computes the rolling average of values forward or backward of the current row within the specified column.

ROLLINGMAX Function

Computes the rolling maximum of values forward or backward of the current row within the specified column. Inputs can be Integer, Decimal, or Datetime.

ROLLINGSTDEV Function

Computes the rolling standard deviation of values forward or backward of the current row within the specified column.

ROLLINGVAR Function

Computes the rolling variance of values forward or backward of the current row within the specified column.

ROLLINGSTDEVSAMP Function

Computes the rolling standard deviation of values forward or backward of the current row within the specified column using the sample statistical method.

ROLLINGVARSAMP Function

Computes the rolling variance of values forward or backward of the current row within the specified column using the sample statistical method.

Also:

Item

Description

MERGE Function

Merges two or more columns of String type to generate output of String type. Optionally, you can insert a delimiter between the merged values.

ROUND Function

Rounds input value to the nearest integer. Input can be an Integer, a Decimal, a column reference, or an expression. Optional second argument can be used to specify the number of digits to which to round.

Source:

In this example, the following data comes from times recorded at regular intervals during a three-lap race around a track. The source data is in cumulative time in seconds (time_sc). You can use ROLLING and other windowing functions to break down the data into more meaningful metrics.

lap

quarter

time_sc

1

0

0.000

1

1

19.554

1

2

39.785

1

3

60.021

2

0

80.950

2

1

101.785

2

2

121.005

2

3

141.185

3

0

162.008

3

1

181.887

3

2

200.945

3

3

220.856

Transformation:

Primary key: Since the quarter information repeats every lap, there is no unique identifier for each row. The following steps create this identifier:

Transformation Name Change column data type lap,quarter String
Transformation Name New formula Single row formula MERGE(['l',lap,'q',quarter]) 'splitId'

Get split times: Use the following transform to break down the splits for each quarter of the race:

Transformation Name New formula Multiple row formula ROUND(time_sc - PREV(time_sc, 1), 3) splitId 'split_time_sc'

Compute rolling computations: You can use the following types of computations to provide rolling metrics on the current and three previous splits:

Transformation Name New formula Multiple row formula ROLLINGAVERAGE(split_time_sc, 3) splitId 'ravg'
Transformation Name New formula Multiple row formula ROLLINGMAX(split_time_sc, 3) splitId 'rmax'
Transformation Name New formula Multiple row formula ROLLINGMIN(split_time_sc, 3) splitId 'rmin'
Transformation Name New formula Multiple row formula ROUND(ROLLINGSTDEV(split_time_sc, 3), 3) splitId 'rstdev'
Transformation Name New formula Multiple row formula ROUND(ROLLINGVAR(split_time_sc, 3), 3) splitId 'rvar'

Compute rolling computations using sample method: These metrics compute the rolling STDEV and VAR on the current and three previous splits using the sample method:

Transformation Name New formula Multiple row formula ROUND(ROLLINGSTDEVSAMP(split_time_sc, 3), 3) splitId 'rstdev_samp'
Transformation Name New formula Multiple row formula ROUND(ROLLINGVARSAMP(split_time_sc, 3), 3) splitId 'rvar_samp'

Results:

When the above transforms have been completed, the results look like the following:

lap

quarter

splitId

time_sc

split_time_sc

rvar_samp

rstdev_samp

rvar

rstdev

rmin

rmax

ravg

1

0

l1q0

0

1

1

l1q1

20.096

20.096

0

0

20.096

20.096

20.096

1

2

l1q2

40.53

20.434

0.229

0.479

0.029

0.169

20.096

20.434

20.265

1

3

l1q3

61.031

20.501

0.154

0.392

0.031

0.177

20.096

20.501

20.344

2

0

l2q0

81.087

20.056

0.315

0.561

0.039

0.198

20.056

20.501

20.272

2

1

l2q1

101.383

20.296

0.142

0.376

0.029

0.17

20.056

20.501

20.322

2

2

l2q2

122.092

20.709

0.617

0.786

0.059

0.242

20.056

20.709

20.39

2

3

l2q3

141.886

19.794

0.621

0.788

0.113

0.337

19.794

20.709

20.214

3

0

l3q0

162.581

20.695

0.579

0.761

0.139

0.373

19.794

20.709

20.373

3

1

l3q1

183.018

20.437

0.443

0.666

0.138

0.371

19.794

20.709

20.409

3

2

l3q2

203.493

20.475

0.537

0.733

0.113

0.336

19.794

20.695

20.35

3

3

l3q3

222.893

19.4

0.520

0.721

0.252

0.502

19.4

20.695

20.252

You can reduce the number of steps by applying awindowtransform such as the following:

Transformation Name Window lap rollingaverage(split_time_sc, 0, 3) rollingmax(split_time_sc, 0, 3) rollingmin(split_time_sc, 0, 3) round(rollingstdev(split_time_sc, 0, 3), 3) round(rollingvar(split_time_sc, 0, 3), 3) round(rollingstdevsamp(split_time_sc, 0, 3), 3) round(rollingvarsamp(split_time_sc, 0, 3), 3) lap lap

However, you must rename all of the generated windowX columns.