Filtering rows based on cascading sequence

[mef47]

Hi there

I have data in the format given below.

Sample ID	Model	S	K
BH00345	mm23	0.029	0.443
BH00345	mm15	0.022	0.018
BH00345	mm14	0.047	1.228
BH00345	gm23	0.04	0.669
BH00345	gm15	0.057	2.709
BH00345	gm14	0.042	0.883
BH00746	mm24	0.113	0.32
BH00746	mm23	0.046	0.11
BH00746	mm15	0.019	0.658
BH00746	gm24	0.023	0.129
BH00746	gm23	0.019	0.127
BH00746	gm15	0.144	0.031
BH00964	mm24	0.175	0.519
BH00964	mm23	0.141	0.513
BH00964	gm24	0.171	0.403
BH00964	gm23	0.167	0.031
BH00583	mm23	0.039	0.029
BH00583	gm23	0.064	0.115

Samples have data for both mm and gm model types and for several iterations of each model as indicated by the model suffix number. 

Newer samples do not contain data for all older models. 

Older samples do not contain data for all newer models. 

I want to condense the table so that the Sample ID column contains unique values, and each row contains data from the most recent mm model. 

eg. for the above table, I would like the output to give

Sample ID	Model	S	K
BH00345	mm23	0.029	0.443
BH00746	mm24	0.113	0.32
BH00964	mm24	0.175	0.519
BH00583	mm23	0.039	0.029

Full data set contains 3554 rows and 747 unique sample ID codes

Thanks for your help

<edit. sorry for the horrendous table!! I cant seem to change the width of the columns>

[v-zhenbw-msft · Accepted solution]

Hi @mef47 ,

We can use the Power Query Editor to meet your requirement.

1. Filter the Model column, delete the rows that contain “gm”.

2. We sort the sample ID Ascending, and sort the Model Descending.

3. Then we need to Group the Sample ID column.

4. And we can add a custom column to add an index column. Each group has an index column.

5. Delete the first two columns. And expand the last column.

6. At last we need to filter the Rank column equal 1, then delete the rank column.

The result table just has four rows.

If it doesn’t meet your requirement, could you please show the exact expected result based on the table that you have shared?

Best regards,

Community Support Team _ zhenbw

If this post helps, then please consider Accept it as the solution to help the other members find it more quickly.

BTW, pbix as attached.

[dedelman_clng]

Hi @mef47  -

Presuming that the "latest" model will be the MM with the highest number, we can create the following 3 measures:

Latest Model = CALCULATE(MAX(Samples[Model]))

Latest K =
VAR __model = [Latest Model]
VAR __sample =
    SELECTEDVALUE ( Samples[Sample ID] )
RETURN
    LOOKUPVALUE (
        Samples[K],
        Samples[Model], __model,
        Samples[Sample ID], __sample
    )

Latest S =
VAR __model = [Latest Model]
VAR __sample =
    SELECTEDVALUE ( Samples[Sample ID] )
RETURN
    LOOKUPVALUE (
        Samples[S],
        Samples[Model], __model,
        Samples[Sample ID], __sample
    )

Create a table visual, put Sample ID as the first value, then the 3 measures:

Hope this helps,

David

[mef47]

Hi David

Not quite what I was after but helpful enough to get me to where I wanted to go.

The ALLEXCEPT function what what I was after

I ended up going with:

LATEST_MODEL = CALCULATE(MAX('Samples'[MODEL]),ALLEXCEPT('Samples','Samples'[Sample ID]))

Thanks