You are on page 1of 6

IF (that's a big IF) I can estimate the carrying capacity K then the technique described seems to work okay.

The "actual" parameters which generate the sigmoid are at J5-J7 (and can be played with).
The deviation from a true sigmoid is at F6 (and can be played with).
Pressing F9 generates another set of "almost sigmoid" data.

ms to work okay.

dN/dx=bN(K-N)/K has as solution: N = K/(1+A exp(-bx)) where A is some constant.


Consider the equivalent K/N - 1 = A exp(-bx) . Now we must find A and b so that A exp(-bx) is a "best fit" for K/N - 1
Instead, we find A and b so that log(A) - bx is a "best fit" for y = log(K/N - 1)
That is, we find the regression line for the set (x,y).
Unfortunately, we need to estimate K beforehand!
"actual" A =
x
N
y = log[K/N-1]
K-Error = E =
0.5
"actual" b =
0.00
0.083
3.558
Press F9 for another trial
"actual" K =

4.00
2.701

0.228

2.25

0.252

2.50

0.289

2.500
3.00
2.391
2.00
2.239

2.75

0.312

3.00

0.334

3.25

0.414

3.50

0.434

3.75

0.473

4.00

0.514

1.777
-1.00
1.674
-2.00
1.576

4.25

0.576

1.436

4.50

0.630

1.325

4.75

0.790

1.028

5.00

0.842

0.942

5.25

0.824

0.971

5.50

0.897

0.852

5.75

1.112

0.530

6.00

1.088

0.564

6.25

1.223

0.373

6.50

1.174

0.442

6.75

1.269

0.311

7.00

1.515

-0.020

7.25

1.636

-0.181

7.50

1.646

-0.196

7.75

1.677

-0.237

8.00

1.650

-0.201

8.25

1.955

-0.627

8.50

1.757

-0.347

8.75

2.108

-0.860

9.00

1.901

-0.548

9.25

2.317

-1.222

9.50

2.057

-0.780

9.75

2.281

-1.155

10.00

2.146

-0.922

10.25

2.226

-1.056

10.50

2.220

-1.046

10.75

2.422

-1.432

11.00

2.155
1.00
2.077
0.00
1.831

actual
datadata
actual

y = -0.4608x + 3.4045

estimated
3.0
2.5
2.0
1.5
1.0
0.5
0.0

Linear Regression

0.189

2.00

Linear (y = log[K/N-1] )

1.75

y = log[K/N-1]

2.819

2.911

0.169

0.155

1.50

1.25

3.006

0.142

hence log(A) = 3.4045

Slope = -0.4608

1.00

and "estimated" A =
hence "estimated" b = 0.46
Best Fit : N = K/{1 + 30.10exp(-0.46t) }
Intercept = 3.4045

12

3.055

10

0.135

3.333

0.75

3.278

0.103

0.109

0.50

0.25

11.25
11.50
11.75
12.00
12.25
12.50
12.75
13.00
13.25
13.50
13.75
14.00
14.25
14.50

s a "best fit" for K/N - 1

30
0.5
3

estimated
0.096
0.108

30.10

0.120
0.135
0.150
0.167

estimated

ctual data

0.187
0.208
0.231
0.257
0.285
0.317
0.351
0.388
0.429

14

13

12

11

10

0.473
0.520
0.572
0.627
0.686
0.749
0.815
0.886
0.959
1.036
1.115
1.197
1.281
1.366
1.452
1.538
1.625
1.710
1.794
1.876
1.956
2.033
2.107
2.177
2.244
2.307
2.367
2.422
2.474
2.522

2.567
2.608
2.645
2.680
2.711
2.740
2.766
2.790
2.811
2.831
2.848
2.864
2.878
2.891

You might also like