How works XYZ classification ?

[Former Member]

Dear APO community

I'm currently implementing the new APO functionality which performs the ABC / XYZ analysis. I was able to simulate easily the expected result in an Excel file for the ABC part but the behavior of the XYZ part remains for me a mystery. I used the variance coefficient as method in the XYZ profil. I get the theorical formula of this coefficient on a website specialized

Coefficient
of variation = Standard deviation (√(((D-D1)²+ (D-D1)² … (D-Dn)²)/n-1) / Mean
average demand (D1+D2+..Dn / n)

I customized the thresholds in the XYZ profil according to the calculated coefficients in my Excel file but I don’t get the expected result  when I perform the analysis.

Does anyone already successfully experiment this XYZ analysis and give me a feedback about the XYZ formula ?

What means the value maintained in XYZ profil ? Upper limit ?

Best Regards

Vincent

Product	Average Demand	Standard   Deviation	Coef of variation	Classif XYZ	S-12	S-11	S-10	S-9	S-8	S-7	S-6	S-5	S-4	S-3	S-2	S-1
P1	10	1,23	0,12	X	10	9	10	11	12	9	10	11	10	9	10	13
P2	29	14,28	0,49	Y	30	20	30	40	50	60	20	15	20	15	30	20
P3	108	150,50	1,39	Z	100		100	300	100		100		500			100
		(D-Dn)²	0	2	0	0	3	2	0	0	0	2	0	7
Nbr of périod	12		1	84	1	117	434	951	84	201	84	201	1	84
69	11736		69	36736	69	11736	69	11736	153403	11736	11736	69
Thresolds XYZ	X	0,3
	Y	0,8
	Z	1,5

[former_member254329 · Accepted solution]

Bonjour Vincent,

I am currently implementing ABCXYZ and it is working fine for me. The formula used in the background for classification is the same as you are describing. You can view the details  through SE80, INCLUDE /SAPAPO/ABCCLASS and METHOD  /sapapo/cl_abc_xyz=>calc_variance

In your data sample, your average formula is converting blank to zero for P3. Please try and see if there would be a difference if you put zeros instead of blanks in DP. You need to be carefull with periodicity values and aggregate level defined in the config.

The value maintained in the XYZ profile corresponds to the upper limit for variance coefficient indeed.

Cheers,

MZ

[Former Member · Accepted solution]

Hello Maxence

Thanks a lot for your reply. I finally found the issue by myself and you're right, the system doesn't consider null values in the formula. I enhanced a badi to change the logic and take into account 0 values.

Best Regards

Vincent

[former_member254329 · Accepted solution]

There is an oss note that now adress this issue 2090091