To ensure that your results are valid, consider the following guidelines when you collect data, perform the analysis, and interpret your results.
- The data must include at least 2 control factors
A designed experiment in Minitab must have at least 2 control factors. A Taguchi analysis treats all factors as categorical, although the actual measurements may be on a continuous scale.
- If you have only one categorical factor and no continuous predictors, use One-Way
- If you have one continuous factor, use Fitted Line
- A dynamic design must have 1 signal factor
- A signal factor has a range of settings that are controlled by the user of the product to make use of its intended function. Signal factors are used in dynamic experiments, in which the response is measured at each level of the signal. The objective is to improve the relationship between the signal factor and the response.
- An example of a signal factor is gas pedal position. The response, the car's speed, should have a consistent relationship with the amount of pressure applied to the gas pedal.
- You must have a minimum of 2 responses
- Structure your data in the worksheet so that each row contains the control factors in the inner array and the response values from one entire run of the noise factors in the outer array. For more information, go to How to arrange Taguchi response data in the worksheet.
- The maximum number of response columns you can enter is 50. Usually, the minimum number of response columns you can enter is 2. However, the minimum number of response columns depends on the design. You can have 1 response only when:
- Your design contains replicates.
- You measure more than one noise factor at each run and create your design so it has multiple runs at each combination of factor settings.
- You are using the Larger is Better or Smaller is Better signal-to-noise ratio, and you do not analyze or store the standard deviation.
- The response variable should be continuous
If the response variable is categorical, your model is less likely to meet the assumptions of the analysis, to accurately describe your data, or to make useful predictions.
- Ensure that the measurement system produces reliable response data
If the variability in your measurement system is too large, your experiment may lack the power to detect important effects.
- Each observation should be independent from all other observations
If your individual observations are dependent, your results might not be valid. Consider the following points to determine whether your observations are independent:
- If an observation provides no information about the value of a different observation, the observations are independent.
- If an observation provides information about a different observation, the observations are dependent.
- Collect data using best practices
To ensure that your results are valid, consider the following guidelines:
- Confirm that the data represent the population of interest.
- Collect enough data to provide the necessary precision.
- Record the data in the order it was collected.
- The model should provide a good fit to the data
If the model does not fit the data, the results can be misleading. In the output, use the residual plots, the diagnostic statistics for unusual observations, and the model summary statistics to determine how well the model fits the data.
To fit a linear model, click Analysis and specify model options when you perform the analysis.