When conducting an orthogonal table experiment, experimental data may not be obtained. (1) In the case of the experiment fails, (2) the products may not be created. If the intended result is not produced as in (2), the SN ratio of the row is set to a value slightly smaller than the smallest SN ratio of other rows. In the case of (1) ,it may have been a good SN ratio if the experiment was completed. Therefore, "treatment of missing data" is performed. Please see the following material because it is complicated.
Click here for data → QE17
p.1: For example, suppose that the SN ratios on the 9th and 10th rows were not obtained (marked with x). (1) Obtain and substitute the average value of the SN ratio excluding the 9th and 10th rows, and draw the graph of factorial effects 1. (2) Regarding the 9th and 10th rows, control factors except the effective control factors D, E, G, and H are set at level 3 (average value of all SN ratios, this time 14.47), and the average value is calculated again. The average value is divided by 11 by adding the SN ratio of the blue frame. The 9th row becomes 14.83, and the 10th row becomes 14.22.
p.2: Substitute this value into the 9th and 10th rows where there were missing data, and draw the graph of factorial effects II. Even if there are missing items, do not give up. Since the orthogonal array does not end once, let's put it in the next orthogonal array experiment if there are any concerning parameters.