ANOVA for Quadratic model

Response 1: transmitan

Source Sum of Squares df Mean Square F-value p-value

Model 0,7178 2 0,3589 9,63 0,0193 significant
⁽¹⁾Linear Mixture 0,6238 1 0,6238 16,75 0,0094
AB 0,0939 1 0,0939 2,52 0,1731
Residual 0,1862 5 0,0372
Lack of Fit 0,1239 2 0,0620 2,98 0,1936 not significant
Pure Error 0,0623 3 0,0208
Cor Total 0,9040 7

⁽¹⁾ Inference for linear mixtures uses Type I sums of squares.

Mixture Component coding is L_Pseudo.

Sum of squares is Type III - Partial

The Model F-value of 9,63 implies the model is significant. There is only a 1,93% chance
that an F-value this large could occur due to noise.

P-values less than 0,0500 indicate model terms are significant. In this case A, B are
significant model terms. Values greater than 0.1000 indicate the model terms are not
significant. If there are many insignificant model terms (not counting those required to
support hierarchy), model reduction may improve your model.

The Lack of Fit F-value of 2,98 implies the Lack of Fit is not significant relative to the pure
error. There is a 19,36% chance that a Lack of Fit F-value this large could occur due to noise.
Non-significant lack of fit is good -- we want the model to fit.

Coefficients in Terms of Coded Factors

Component Coefficient Estimate df Standard Error 95% CI Low 95% CI High VIF
A-tween 80 99,56 1 0,1319 99,22 99,90 1,46
B-lesitin 98,82 1 0,1319 98,48 99,15 1,46
AB -0,9711 1 0,6115 -2,54 0,6008 1,96

The coefficient estimate represents the expected change in response per unit change in factor
value when all remaining factors are held constant. The intercept in an orthogonal design is
the overall average response of all the runs. The coefficients are adjustments around that
average based on the factor settings. When the factors are orthogonal the VIFs are 1; VIFs
greater than 1 indicate multi-colinearity, the higher the VIF the more severe the correlation of
factors. As a rough rule, VIFs less than 10 are tolerable.

Fit Statistics

Std. Dev. 0,1930 R² 0,7940

Mean 99,08 Adjusted R² 0,7116
C.V. % 0,1948 Predicted R² 0,5084
Adeq Precision 6,3007

The Predicted R² of 0,5084 is not as close to the Adjusted R² of 0,7116 as one might
normally expect; i.e. the difference is more than 0.2. This may indicate a large block effect or
a possible problem with your model and/or data. Things to consider are model reduction,
response transformation, outliers, etc. All empirical models should be tested by doing
confirmation runs.

Adeq Precision measures the signal to noise ratio. A ratio greater than 4 is desirable. Your
ratio of 6,301 indicates an adequate signal. This model can be used to navigate the design
space.

Final Equation in Terms of Real Components

transmitan =
+100,66086 tween 80
+105,45129 lesitin
-15,53820 tween 80 * lesitin

Final Equation in Terms of L_Pseudo Components

transmitan =
+99,56 A
+98,82 B
-0,9711 AB

The equation in terms of coded factors can be used to make predictions about the response for
given levels of each factor. By default, the high levels of the mixture components are coded
as +1 and the low levels are coded as 0. The coded equation is useful for identifying the
relative impact of the factors by comparing the factor coefficients.
Final Equation in Terms of Actual Components

transmitan =
+25,16522 tween 80
+26,36282 lesitin
-0,971137 tween 80 * lesitin

The equation in terms of actual factors can be used to make predictions about the response for
given levels of each factor. Here, the levels should be specified in the original units for each
factor. This equation should not be used to determine the relative impact of each factor
because the coefficients are scaled to accommodate the units of each factor and the intercept
is not at the center of the design space.

ANOVA for Quartic model

Response 2: viskositas

Source Sum of Squares df Mean Square F-value p-value

Model 10,05 4 2,51 953,68 < 0.0001 significant
⁽¹⁾Linear Mixture 6,61 1 6,61 2511,14 < 0.0001
AB 1,80 1 1,80 684,26 0,0001
AB(A-B) 0,6588 1 0,6588 250,18 0,0005
AB(A-B)² 0,1290 1 0,1290 48,99 0,0060
Pure Error 0,0079 3 0,0026
Cor Total 10,05 7

⁽¹⁾ Inference for linear mixtures uses Type I sums of squares.

Mixture Component coding is L_Pseudo.

Sum of squares is Type III - Partial

The Model F-value of 953,68 implies the model is significant. There is only a 0,01% chance
that an F-value this large could occur due to noise.

P-values less than 0,0500 indicate model terms are significant. In this case A, B, AB, AB(A-
B), AB(A-B)² are significant model terms. Values greater than 0.1000 indicate the model
terms are not significant. If there are many insignificant model terms (not counting those
required to support hierarchy), model reduction may improve your model.

Fit Statistics

Std. Dev. 0,0513 R² 0,9992

Mean 3,18 Adjusted R² 0,9982
C.V. % 1,61 Predicted R² NA⁽¹⁾
Adeq Precision 66,4302

⁽¹⁾ Case(s) with leverage of 1.0000: Pred R² and PRESS statistic not defined.

Adeq Precision measures the signal to noise ratio. A ratio greater than 4 is desirable. Your
ratio of 66,430 indicates an adequate signal. This model can be used to navigate the design
space.

Coefficients in Terms of Coded Factors

Component Coefficient Estimate df Standard Error 95% CI Low 95% CI High VIF
A-tween 80 5,11 1 0,0363 5,00 5,23 1,56
B-lesitin 2,42 1 0,0363 2,30 2,54 1,56
AB -4,65 1 0,1778 -5,22 -4,08 2,34
AB(A-B) -6,49 1 0,4105 -7,80 -5,19 1,13
AB(A-B)² -6,84 1 0,9773 -9,95 -3,73 1,59
The coefficient estimate represents the expected change in response per unit change in factor
value when all remaining factors are held constant. The intercept in an orthogonal design is
the overall average response of all the runs. The coefficients are adjustments around that
average based on the factor settings. When the factors are orthogonal the VIFs are 1; VIFs
greater than 1 indicate multi-colinearity, the higher the VIF the more severe the correlation of
factors. As a rough rule, VIFs less than 10 are tolerable.

Final Equation in Terms of L_Pseudo Components

viskositas =
+5,11 A
+2,42 B
-4,65 AB
-6,49 AB(A-B)
-6,84 AB(A-B)²

The equation in terms of coded factors can be used to make predictions about the response for
given levels of each factor. By default, the high levels of the mixture components are coded
as +1 and the low levels are coded as 0. The coded equation is useful for identifying the
relative impact of the factors by comparing the factor coefficients.

Final Equation in Terms of Real Components

viskositas =
+40,21000 tween 80
+1850,07000 lesitin
-3719,20000 tween 80 * lesitin
+3086,50667 tween 80 * lesitin * (tween 80-lesitin)
-1751,04000 tween 80 * lesitin * (tween 80-lesitin)²

Final Equation in Terms of Actual Components

viskositas =
+10,05250 tween 80
+462,51750 lesitin
-232,45000 tween 80 * lesitin
+48,22667 tween 80 * lesitin * (tween 80-lesitin)
-6,84000 tween 80 * lesitin * (tween 80-lesitin)²

The equation in terms of actual factors can be used to make predictions about the response for
given levels of each factor. Here, the levels should be specified in the original units for each
factor. This equation should not be used to determine the relative impact of each factor
because the coefficients are scaled to accommodate the units of each factor and the intercept
is not at the center of the design space.

ANOVA for Cubic model

Response 3: %inhibisi

Source Sum of Squares df Mean Square F-value p-value

Model 535,48 3 178,49 14,63 0,0127 significant
⁽¹⁾Linear Mixture 37,12 1 37,12 3,04 0,1560
AB 222,68 1 222,68 18,26 0,0129
AB(A-B) 275,67 1 275,67 22,60 0,0089
Residual 48,79 4 12,20
Lack of Fit 11,86 1 11,86 0,9630 0,3988 not significant
Pure Error 36,93 3 12,31
Cor Total 584,27 7

⁽¹⁾ Inference for linear mixtures uses Type I sums of squares.

Mixture Component coding is L_Pseudo.

Sum of squares is Type III - Partial

The Model F-value of 14,63 implies the model is significant. There is only a 1,27% chance
that an F-value this large could occur due to noise.

P-values less than 0,0500 indicate model terms are significant. In this case A, B, AB, AB(A-
B) are significant model terms. Values greater than 0.1000 indicate the model terms are not
significant. If there are many insignificant model terms (not counting those required to
support hierarchy), model reduction may improve your model.

The Lack of Fit F-value of 0,96 implies the Lack of Fit is not significant relative to the pure
error. There is a 39,88% chance that a Lack of Fit F-value this large could occur due to noise.
Non-significant lack of fit is good -- we want the model to fit.

Fit Statistics

Std. Dev. 3,49 R² 0,9165

Mean 36,19 Adjusted R² 0,8539
C.V. % 9,65 Predicted R² 0,6291
Adeq Precision 10,1273

The Predicted R² of 0,6291 is not as close to the Adjusted R² of 0,8539 as one might
normally expect; i.e. the difference is more than 0.2. This may indicate a large block effect or
a possible problem with your model and/or data. Things to consider are model reduction,
response transformation, outliers, etc. All empirical models should be tested by doing
confirmation runs.

Adeq Precision measures the signal to noise ratio. A ratio greater than 4 is desirable. Your
ratio of 10,127 indicates an adequate signal. This model can be used to navigate the design
space.

Coefficients in Terms of Coded Factors

Component Coefficient Estimate df Standard Error 95% CI Low 95% CI High VIF
A-tween 80 41,46 1 2,46 34,64 48,29 1,55
B-lesitin 41,25 1 2,46 34,43 48,08 1,55
AB -47,28 1 11,07 -78,01 -16,56 1,96
AB(A-B) 132,83 1 27,94 55,25 210,40 1,12
The coefficient estimate represents the expected change in response per unit change in factor
value when all remaining factors are held constant. The intercept in an orthogonal design is
the overall average response of all the runs. The coefficients are adjustments around that
average based on the factor settings. When the factors are orthogonal the VIFs are 1; VIFs
greater than 1 indicate multi-colinearity, the higher the VIF the more severe the correlation of
factors. As a rough rule, VIFs less than 10 are tolerable.

Final Equation in Terms of L_Pseudo Components

%inhibisi =
+41,46 A
+41,25 B
-47,28 AB
+132,83 AB(A-B)

The equation in terms of coded factors can be used to make predictions about the response for
given levels of each factor. By default, the high levels of the mixture components are coded
as +1 and the low levels are coded as 0. The coded equation is useful for identifying the
relative impact of the factors by comparing the factor coefficients.

Final Equation in Terms of Real Components

%inhibisi =
-122,20990 tween 80
+7427,84265 lesitin
-13507,86510 tween 80 * lesitin
+8500,90667 tween 80 * lesitin * (tween 80-lesitin)

Final Equation in Terms of Actual Components

%inhibisi =
-30,55248 tween 80
+1856,96066 lesitin
-844,24157 tween 80 * lesitin
+132,82667 tween 80 * lesitin * (tween 80-lesitin)

The equation in terms of actual factors can be used to make predictions about the response for
given levels of each factor. Here, the levels should be specified in the original units for each
factor. This equation should not be used to determine the relative impact of each factor
because the coefficients are scaled to accommodate the units of each factor and the intercept
is not at the center of the design space.

Constraints
 Name Goal Lower Limit Upper Limit Lower Weight Upper Weight Importance
A:tween 80 is in range 2,5 3,5 1 1 3
B:lesitin is in range 0,5 1,5 1 1 3
transmitan maximize 90 99 1 1 4
viskositas minimize 1 200 1 1 3
%inhibisi maximize 18,06 44,96 1 1 5

Solutions

2 Solutions found

Number tween 80 lesitin transmitan viskositas %inhibisi Desirability

1 3,249 0,751 99,191 2,638 44,960 0,998 Selected
2 2,500 1,500 98,816 2,420 41,254 0,932

Starting Points

Number of Starting Points: 105

tween 80 lesitin
3,25 0,75
3 1
3,5 0,5
2,75 1,25
2,5 1,5
3,37585 0,62415
3,07714 0,922864
2,92292 1,07708
2,68117 1,31883
3,02155 0,97845
2,76426 1,23574
2,86125 1,13875
2,90441 1,09559
3,30286 0,697137
3,47075 0,529249
3,03414 0,965856
3,28769 0,712314
3,0861 0,913903
2,65857 1,34143
2,8574 1,1426
2,6814 1,3186
2,88474 1,11526
3,07308 0,926921
2,8399 1,1601
3,09692 0,903083
3,05259 0,947412
3,30873 0,691266
2,55783 1,44217
3,19112 0,80888
3,063 0,937001
2,8402 1,1598
2,53433 1,46567
3,17754 0,822463
2,98771 1,01229
3,36359 0,636414
3,15803 0,841969
2,6057 1,3943
2,6149 1,3851
3,42697 0,57303
2,72036 1,27964
3,12688 0,873116
3,07986 0,92014
3,14027 0,859735
3,36685 0,633148
2,84971 1,15029
3,4962 0,503797
3,24069 0,759308
2,97178 1,02822
3,11223 0,887773
2,52141 1,47859
2,8884 1,1116
2,5853 1,4147
3,4659 0,534103
2,66125 1,33875
2,61013 1,38987
2,82463 1,17537
3,25868 0,741317
2,64423 1,35577
3,44744 0,552559
2,60086 1,39914
3,38838 0,611618
3,05585 0,944149
3,4845 0,515498
2,75334 1,24666
2,99568 1,00432
2,75092 1,24908
2,88053 1,11947
3,38386 0,616136
3,14497 0,855027
2,58108 1,41892
2,66611 1,33389
3,05732 0,942679
2,51756 1,48244
2,62979 1,37021
3,13006 0,86994
3,30027 0,699726
2,85055 1,14945
2,78592 1,21408
2,67623 1,32377
3,38929 0,61071
3,42991 0,570087
2,92821 1,07179
3,38031 0,619691
3,22781 0,772186
2,60634 1,39366
2,97678 1,02322
3,02693 0,973066
3,15042 0,849582
3,41568 0,584321
2,87651 1,12349
2,87785 1,12215
3,31708 0,682919
3,46338 0,536617
2,66114 1,33886
2,56314 1,43686
3,26977 0,730228
3,08307 0,916926
3,38961 0,61039
3,10054 0,899457
3,42089 0,579108
3,0446 0,9554
3,36021 0,639793
2,88627 1,11373
3,43282 0,567177
3,04979 0,950213
Factors

Component Name Level Low Level High Level Std. Dev. Coding
A tween 80 3,25 2,50 3,50 0,0000 Actual
B lesitin 0,7509 0,5000 1,50 0,0000 Actual
Total = 4,00

Point Prediction
Two-sided    Confidence = 95%    Population = 99%

Solution 1 95% CI 95% CI 95% TI 95% TI

Predicted Predicted
of 2 Observed Std Dev SE Mean low for high for low for high for
Mean Median
Response Mean Mean 99% Pop 99% Pop
transmitan 99,1908 99,1908 0,193 0,0949734 98,9466 99,4349 97,9607 100,421
viskositas 2,63766 2,63766 0,051316 0,0511544 2,47486 2,80045 2,1313 3,14401
%inhibisi 44,9603 44,9603 3,49249 2,89212 36,9305 52,9901 17,4529 72,4677

Confirmation Location #1

tween 80 lesitin
3,44857 0,551435
transmitan viskositas %inhibisi

Confirmation

Two-sided    Confidence = 95%

Solution 1 of 2 Predicted Predicted 95% PI Data 95% PI

Observed Std Dev n SE Pred
Response Mean Median low Mean high
transmitan 99,4745 99,4745 0,193 1 0,223472 98,9 100,049
viskositas 4,19671 4,19671 0,051316 1 0,0679316 3,98052 4,4129
%inhibisi 44,9602 44,9602 3,49249 1 4,10418 33,5651 56,3552

Factors

Component Name Level Low Level High Level Std. Dev. Coding
A tween 80 3,25 2,50 3,50 0,0000 Actual
B lesitin 0,7509 0,5000 1,50 0,0000 Actual
Total = 4,00

Confirmation

Two-sided    Confidence = 95%

Solution 1 of 2 Predicted Predicted 95% PI Data 95% PI

Observed Std Dev n SE Pred
Response Mean Median low Mean high
transmitan 99,4745 99,4745 0,193 1 0,223472 98,9 100,049
 viskositas 4,19671 4,19671 0,051316 1 0,0679316 3,98052 4,4129
%inhibisi 44,9602 44,9602 3,49249 1 4,10418 33,5651 56,3552

A B AB AB(A-B) AB(A-B)²
transmitan 99,5602 98,8155 -0,971137
p-values 0,0094 0,0094 0,1731
viskositas 5,115 2,42 -4,65 -6,49333 -6,84
p-values < 0.0001 < 0.0001 0,0001 0,0005 0,0060
%inhibisi 41,4639 41,2539 -47,2816 132,827
p-values 0,1560 0,1560 0,0129 0,0089

Design Constraints

Mixture Coding: Actual

Low Limit Constraint High Limit

2,500 ≤ A:tween 80 ≤ 3,500
0,500 ≤ B:lesitin ≤ 1,500
A+B = 4,000