cancel
Showing results for
Did you mean: Frequent Visitor

## Exponential Moving Average in Power Query M ?

Hello! How can an exponential moving average be created in Power Query M?  I already know how to do this in Excel, and DAX can be a pain in the tail.  Here's a table of example data.  Thank you!

Date Value
1/1/2021 0.888073856
1/2/2021 0.087243733
1/3/2021 0.597531606
1/4/2021 0.25492501
1/5/2021 0.836644673
1/6/2021 0.253785258
1/7/2021 0.757333787
1/8/2021 0.958886707
1/9/2021 0.598108137
1/10/2021 0.180565909
1/11/2021 0.788755716
1/12/2021 0.888126908
1/13/2021 0.231740519
1/14/2021 0.107410191
1/15/2021 0.648458581
1/16/2021 0.582213204
1/17/2021 0.191300599
1/18/2021 0.864502562
1/19/2021 0.583108477

3 ACCEPTED SOLUTIONS  Super User IV

@jsflint

Can you share the expected value based on the sample data that you have given?

You can save the  Excel file with the results in One Drive or Google Drive and share the link

Proud to be a Super User! Frequent Visitor

Here it is  Super User IV

@jsflint

Please find attached the PBIX file below my signature with the desired results. Basically, the solution involves a recursive operation to calculate the EMA ( Exponential Moving Average in Power Query ).

There are two columns, Average and EMA. Complete M Code:

``````let
Source = Table.FromRows(
Json.Document(
Binary.Decompress(
Binary.FromText(
BinaryEncoding.Base64
),
Compression.Deflate
)
),
let
_t = ((type nullable text) meta [Serialized.Text = true])
in
type table [Date = _t, Value = _t]
),
#"Changed Type" = Table.TransformColumnTypes(
Source,
{{"Date", type date}, {"Value", type number}}
),
Step1 = Table.AddIndexColumn(#"Changed Type", "Index", 0, 1, Int64.Type),
Step1,
"Average",
each if [Index] > 6 then List.Average(List.Range(Step1[Value], _[Index] - 7, 7)) else null
, type number),
Step2,
"EMA",
each
if [Index] > 6 then
let
start = List.First(List.RemoveNulls(Step2[Average])),
vlist = List.Range(Step2[Value], 8, _[Index] - 7),
acc = List.Accumulate(
vlist,
start,
(state, current) => (current - state) * (2 / (1 + 7)) + state
)
in
if [Index] = 7 then start else acc
else
null,
type number )
in
Step3``````

Proud to be a Super User!

4 REPLIES 4  Super User IV

@jsflint

Can you share the expected value based on the sample data that you have given?

You can save the  Excel file with the results in One Drive or Google Drive and share the link

Proud to be a Super User! Frequent Visitor

Here it is  Super User IV

@jsflint

Please find attached the PBIX file below my signature with the desired results. Basically, the solution involves a recursive operation to calculate the EMA ( Exponential Moving Average in Power Query ).

There are two columns, Average and EMA. Complete M Code:

``````let
Source = Table.FromRows(
Json.Document(
Binary.Decompress(
Binary.FromText(
BinaryEncoding.Base64
),
Compression.Deflate
)
),
let
_t = ((type nullable text) meta [Serialized.Text = true])
in
type table [Date = _t, Value = _t]
),
#"Changed Type" = Table.TransformColumnTypes(
Source,
{{"Date", type date}, {"Value", type number}}
),
Step1 = Table.AddIndexColumn(#"Changed Type", "Index", 0, 1, Int64.Type),
Step1,
"Average",
each if [Index] > 6 then List.Average(List.Range(Step1[Value], _[Index] - 7, 7)) else null
, type number),
Step2,
"EMA",
each
if [Index] > 6 then
let
start = List.First(List.RemoveNulls(Step2[Average])),
vlist = List.Range(Step2[Value], 8, _[Index] - 7),
acc = List.Accumulate(
vlist,
start,
(state, current) => (current - state) * (2 / (1 + 7)) + state
)
in
if [Index] = 7 then start else acc
else
null,
type number )
in
Step3``````

Proud to be a Super User! Frequent Visitor

Hey @Fowmy , first of all the code you sent works awesome!  However, I've been ultimately trying to create a Triple Exponential Moving Average, and I figure I could just reuse a good part of the code to create a second and third EMA then calculate a final column to create a TEMA.  It's not working out for me right now, and I'd greatly appreciate your assistance again.  I've attached another sample Excel file that includes date, value, average, ema1,  ema2,  ema3, and Tema columns.  The formulas are present, and hopefully makes sense.  Thanks again!!! Announcements #### Welcome to the User Group Public Preview  