Results+of+experiment

The results were been analysed by the R project. The response surface methodology- 2k Factorial design was used for the completion of the analysis.
 * Screening & Optimisation Experiment Results **
 * **Screening experiment results**

According to the results, **sweetness** was the only attribute which seem to be non-significantly accepted. Therefore, the mean values (calculated from the software Compusense® five) were also investigated showing that samples in upper percentage of sugar got higher score than samples with lower limit. As a result, 10% of sugar was considered as the most acceptable for using it in the optimisation (keep it constant). In addition, as for thickness attribute, the response surface results showed that assessors prefer high level of water and consequently a thinner product. Nevertheless, according to the technical specifications and by the legislation, the solid content should be 8-9.5%, so the highest amount of water used in the final product should not exceed the 20%. Thus, it was decided also to keep it constant in 20% in the optimisation procedure.

The results for the rest two factors, **% of** **kiwi juice and cow’s yogurt**, indicated that assessors prefer both of them in low levels .For that reason, it was decided to be examined in three levels in the optimisation experiment. Since there is a fraction on the lower limit of kiwi juice that can be used (according to technical specifications), the lowest limit for the optimisation experiment set as 8%, followed by 10% and 12%. The equivalent levels for the percentage of cow’s yogurt were set to 7.2% (lower), 14.7% (medium) and 22.2% (high).



The results were been analysed by the R project. The response surface methodology- 3k Factorial design was used for the completion of the analysis. Three different models were used to determine the optimal region for both kiwi and milk flavour as following: i) FO (First order) ii) FO (First order) + TWI (Two way interaction), iii) SO (second order). The best model for each factor was chosen by checking the p value and lack of fit. In addition, plots were plotted by using the best model followed by analysis for the best region. The best model for **kiwi flavour** was found to be SO (second order). By using this model, the 3D graphs were plotted to find out the interaction between percentage of cow’s yogurt and percentage of kiwi juice. The 3D contour plots indicated that the higher score achieved for the kiwi flavour was between 6.6-6.7. Nevertheless, according to the standards, the percentage of cow’s yogurt and kiwi juice can’t be reduced beyond the lower limit. Hence, the optimal region for cow’s yogurt and kiwi juice is considered to be 14.5% and 9% respectively. As for the **milk flavour**, SO (second order) model seemed to be the best and therefore. Plots indicated that the optimal region for milk flavour is the lower percentage of kiwi juice and medium percentage of cow’s yogurt. The higher score gained for milk flavour was between 6.6-6.8. Plots strongly show that there is definite interaction between percentage of cow’s yogurt and kiwi juice with respect to the milk flavour. The analysis of **overall acceptability** had also been conducted in order the final decision over optimal region for kiwi flavour and milk flavour to be taken. FO (First order) + TWI (Two way interaction) was the best model. Plots strongly suggest that the optimal region can be achieved with two different combinations, a) with low percentage of kiwi juice and high percentage of cow’s yogurt b) low percentage of cow’s yogurt and high percentage of kiwi juice. The higher score was between 6.5-6.7. There is interaction between percentage cows yogurt and percentage kiwi juice.
 * **Optimisation experiment results**



Back to home page Back to Stage III