Task+5

The response variables that were measured for the product and process optimisation were creaminess, iciness and softness, these are sensory attributes related to the textural properties of ice cream. the purpose of the optimisation was to enhance the texture and quality of our product. To determine the acceptability of our eight ice cream samples, the samples were presented to a panel of 56 untrained assessors. The samples were prepared and presented to the assessors in one session and they were instructed to rate the acceptability of each sample using a 9-point hedonic scale. Samples were prepared on same day as the test, in an identical manner, and were prepared out of sight from the panel members. Products were served frozen, consent forms were given to, and signed by the assessors. Assessors worked independently in partitioned sensory booths and no discussion took place during and after the session. Computerized testing was done using Compusense® five. For all the attributes, Tukey's test for additivity was first conducted to determine if there was any interaction between the samples and assessors and if there was a significant interaction, a Friedman two-way analysis of variance (ANOVA) was conducted, if there was no significant interaction, one factor within subject ANOVA was conducted. If either the Friedman two-way ANOVA or one factor within subject ANOVA test showed a significant difference in the preference of samples, a turkey honestly significant difference by ranks test was conducted. For the one factor within subject ANOVA, A two-sided hypothesis test for the mean values of each paired comparison for the samples factor can be defined such that: H0: μA= μB= μC = μD= μE= μF= μG= μH and H1: μA ≠μB. ≠μC≠μD≠ μE≠ μF≠ μG≠ μH For the Friedman two-way ANOVA test, a two sided alternative hypothesis (H1) can be defined such that: H0 : ∑RA =∑RB =∑RC=∑RD=∑RE =∑RF =∑RG=∑RH and H1 : not all ∑Ri are equal Table: product formulation Table: means obtained from the preference tests for the attribute creaminess, tukey additivity test was significant (0.0002271), this means that there was an interaction between assessors and samples factors. The friedman test showed that there was no significant difference (0.3996) in the preference of all the samples. for the attribute creaminess, tukey additivity test was not significant (0.2332). the one factor within subject ANOVA was conducted and it showed that there was no significant difference (0.1388) in the preference of all the samples. The assessors effect showed there was a significant difference(<2e-16 ***) in the responses of the assessors, this is expected as the each assessor has difference in their preferences. for the attribute iciness, tukey additivity test was not significant (0.4795). the one factor within subject ANOVA was conducted and it showed that there was no significant difference (0.3781) in the preference of all the samples. The assessors effect showed there was a significant difference(<2e-16 ***) in the responses of the assessors, this is expected as the each assessor has difference in their preferences. For overall preference test, tukey additivity test was significant (0.001448), this means that there was an interaction between assessors and samples factors. The friedman test showed that there was no significant difference (0.4055) in the preference of all the samples. For both iciness and softness attributes, The Normal Q-Q plot suggests that the residuals which mostly fall along the straight line. This indicates that the residual term is normally distributed. In the plot of residuals versus fitted values, the red line is almost a straight line on the plot which also suggests that the assumption of a linear regression model is adequate. The plot of standardized residuals versus factor level combinations is also randomly distributed in the same manner regardless of the factor level showing no inequality of variances. Some of the residuals are highlighted which indicate that they may be outliers which might have affected the model fit. These outliers can be removed but removing these suspected outliers affects the balance, since they are few and do not seem to be too far from the line, it can be assumed that they do not influence the model fit. Since theone factor within subject ANOVA was not significant and from the plot, the assumption of a linear regression model is adequate, no further analysis was carried out. The analysis of the product acceptability and preference test data obtained from the sensory tests indicates that there is no significant difference in the preference of all eight samples. For further optimization, product G was selected because it had high mean values.
 * Factors || Product A || Product B || Product C || Product D || Product E || Product F || Product G || Product H ||
 * Carageenan (g) || 0.25 || 0.25 || 0.1 || 0.1 || 0.25 || 0.25 || 0.1 || 0.1 ||
 * Guar gum (g) || 1 || 1 || 1 || 1 || 0.5 || 0.5 || 0.5 || 0.5 ||
 * Panadan (g) || 1 || 0.5 || 1 || 0.5 || 1 || 0.5 || 1 || 0.5 ||
 * Products || Creaminess || Softness || Iciness || Overall preference ||
 * Product A || 6.9 || 6.9 || 6.6 || 6.9 ||
 * Product B || 7.1 || 6.9 || 6.7 || 7.2 ||
 * Product C || 6.9 || 6.9 || 6.5 || 6.8 ||
 * Product D || 7.2 || 6.8 || 6.4 || 6.9 ||
 * Product E || 6.7 || 6.5 || 6.5 || 6.8 ||
 * Product F || 6.9 || 7 || 6.8 || 7.1 ||
 * Product G || 6.9 || 6.9 || 6.9 || 7.2 ||
 * Product H || 7.1 || 7.12 || 6.6 || 7.1 ||