topic Linear Plateau code assistance needed in New SAS User

Linear Plateau code assistance needed

Trwhite — Tue, 12 Jan 2021 20:54:09 GMT

I have been instructed to use a linear plateau for the data below. I am confused with the overall coding linear plateaus that I have found here and other sources online. I hope if I see it used with some of my own data, I will be able to make better sense of the code and apply it to my remaining data sets. Any help would be greatly appreciated. Thank you.

The data is looking at relative crop yields after four levels of crop destruction were applied.

Data:

X= Severity

Y= RelYield

X	Y
0	84.364
0	107.463
0	90.044
0	118.130
25	80.726
25	100.291
25	88.546
25	87.870
50	76.977
50	76.065
50	82.529
50	67.358
75	76.575
75	82.847
75	103.377
75	81.920
0	92.807
0	133.470
0	94.579
0	79.145
25	109.281
25	98.694
25	140.440
25	110.404
50	94.516
50	93.878
50	90.349
50	92.902
75	66.631
75	82.134
75	94.602
75	64.348
0	105.355
0	104.519
0	98.055
0	92.071
25	98.167
25	142.057
25	121.168
25	94.243
50	89.909
50	99.378
50	140.359
50	106.142
75	71.665
75	86.498
75	80.806
75	99.886

Re: Linear Plateau code assistance needed

Reeza — Tue, 12 Jan 2021 21:06:39 GMT

What is your definition of a 'linear plateau'?

Assuming you mean where does the curve flatten, do you remember the first derivatives concepts from calculus?

Where the first derivate is close to zero is likely what you're trying to calculate?

From the data shown what would you expect as output? What would you like to get back if that was your input exactly?

@Trwhite wrote:

I have been instructed to use a linear plateau for the data below. I am confused with the overall coding linear plateaus that I have found here and other sources online. I hope if I see it used with some of my own data, I will be able to make better sense of the code and apply it to my remaining data sets. Any help would be greatly appreciated. Thank you.

The data is looking at relative crop yields after four levels of crop destruction were applied.

Data:

X= Severity

Y= RelYield

X Y

0 84.364

0 107.463

0 90.044

0 118.130

25 80.726

25 100.291

25 88.546

25 87.870

50 76.977

50 76.065

50 82.529

50 67.358

75 76.575

75 82.847

75 103.377

75 81.920

0 92.807

0 133.470

0 94.579

0 79.145

25 109.281

25 98.694

25 140.440

25 110.404

50 94.516

50 93.878

50 90.349

50 92.902

75 66.631

75 82.134

75 94.602

75 64.348

0 105.355

0 104.519

0 98.055

0 92.071

25 98.167

25 142.057

25 121.168

25 94.243

50 89.909

50 99.378

50 140.359

50 106.142

75 71.665

75 86.498

75 80.806

75 99.886

Re: Linear Plateau code assistance needed

Trwhite — Wed, 13 Jan 2021 14:30:08 GMT

Yes, I am hoping to see where the curve flattens on a graph as well as numerically. Thank you for the provided link, it served as a nice review. I believe the yields for severity levels 50 and 75% will display where the line flattens out from the data provided.

Re: Linear Plateau code assistance needed

Rick_SAS — Thu, 14 Jan 2021 19:27:01 GMT

A "linear plateau" usually means that the explanatory variable is measured on a continuous scale ... or at least at more than four points. When the X variable is measured at many locations, you can use a "segmented regression" to estimate the point at which the data flattens out. An example is shown in this article on segmented regression in SAS.

Personally, I don't think these data indicate a plateau. Graphing the data does not indicate a plateau:

proc sgplot data=Have;
   vbox y / category=x connect=mean;
run;