Learn more in our free online course:
Statistical Thinking for Industrial Problem Solving
In this video, we show how to hide and exclude observations from an analysis in JMP. When you hide observations, you remove the data from graphical displays. When you exclude observations, you remove the values from future analyses or calculations.
We’ll use the Temperature data set, which has 100 temperature readings for tools in a machining process. The temperature is tightly controlled. For this process, the ideal operating range is 50 to 55 degrees F.
To start, we create a distribution of Temperature values using the Analyze, Distribution platform.
The values are centered around 52.5, but you can see that there is one value that is extreme relative to the operating range. The temperature for this observation is 64.
We know that this is not a feasible value because the equipment does not function at this temperature. There is likely a data entry error, and you decide to remove the observation from the analysis.
To hide the observation from the graph, we select the point, right-click, and select Row Hide. In the data table, this observation now has a mask as a row state, indicating that the observation is hidden.
Hiding removes the observation from the box plot but not from the histogram. And the observation is still included in the analysis.
To exclude the observation, we can right-click on the row in the data table and select Exclude/Unexclude. Or we select the option from the Rows menu.
The row state for this observation now has a “don’t” symbol, telling JMP not to include this observation in future analyses.
Notice that the observation is still included in the histogram and the summary statistics. We can re-run the analysis and this observation will be excluded.
A simple way to re-run the analysis is to use the red triangle next to Distributions and select Redo and then Redo Analysis.
Note that, instead of hiding and excluding the observation in two steps, we can hide and exclude in one step.
Let’s take a look at the distribution without this observation. The distribution, and the summary statistics, are now what we would expect to see given our knowledge of this process.
