Chapter 3 - Developing the Plan (Part 2) Chapter 4 - Data Step
ME 202
Fall 2011
ME 202 (Fall 2011)
1/ 10
1 / 10
Introduction
Developing the Plan
Recall that in the plan stage, we would like to determine procedures for carrying out the study and collecting the data. We looked at the following sub-stages in past lectures:
Dening the study units and population Determining what variates will be measured Specifying the sampling protocol
We will now address the nal two sub-stages of the plan:
Dening and assessing the measurement systems Planning for data collection
ME 202 (Fall 2011)
2/ 10
2 / 10
Dening and assessing the measurement systems
The measuring process
A measurement system or process is the combination of instruments (or gauges), methods, material, environment and people/operators that are used to produce the measured value of a variate on a sampled unit. Denition: Measurement error is the difference between the measured value and the true value of a variate. We can use a separate PPDAC cycle to investigate the measurement system by itself:
A unit is the act of taking a single measurement. The response variate is a measured value. Possible attributes of interest include the average measurement error or standard deviation (or diversity) of the measurement error.
ME 202 (Fall 2011)
3/ 10
3 / 10
Dening and assessing the measurement systems
Example 1.2 of Chapter 1 Revisited
Recall the example where a micrometer is used to measure the diameter of a ground shaft. The following is a shbone diagram for the measurement system.
ME 202 (Fall 2011)
4/ 10
4 / 10
Planning for data collection
Preparing for the data stage
Address various questions regarding the collection of the data. For example, Who will collect the sample and make the measurements? What might go wrong? What can we do to avoid this? When will the sample be collected and the measurements made? How will we record the data in an effective and efcient manner?
ME 202 (Fall 2011)
5/ 10
5 / 10
The Data Stage
The data stage
This stage may be costly and time-consuming. As such, it is important to continuously monitor the data collection process to ensure that everything goes according to the plan. At all points in this stage, we must record any departures from the plan. Steps:
Execute the plan. Monitor the data as it is collected. Examine the data. Store the data.
ME 202 (Fall 2011)
6/ 10
6 / 10
The Data Stage
Execute the plan
If the plan is experimental, set the explanatory variate values. Select the sample of units according to the sampling protocol and record any departures. Measure variates on each unit.
Record the measurer. Record the gauge. Record time/order of each measurement. Follow measurement systems protocols. Distinguish complete, incomplete, and missing measurements.
ME 202 (Fall 2011)
7/ 10
7 / 10
The Data Stage
Monitor the data collection
Within each sampled unit,
are there any missing values? (You can represent missing data values using "NA" or "NaN". It is not recommended that you represent missing values using numerical values, e.g. 0, 99, etc.) are there any illegal data types? (e.g. letter values instead of numerical values) is the range of data values logically consistent? (e.g. negative values for a variate which is always positive)
Across the units, check for unexpected relationships
between measurements and time of measurement. measurements and the gauge used. between measurements and the measurer.
ME 202 (Fall 2011)
8/ 10
8 / 10
The Data Stage
Examine the data
Conduct visual and graphical checks to look for patterns of internal consistency. Is the data logically consistent with any prior information? Look for extreme observations (sometimes called outliers) which seem to be away from the bulk of the data. We can perform, for example, a marginal examination (look a one variate at a time), a joint examination (look at multiple variates at once) or a conditional examination (look at one or multiple variates which are organized according to the value of another variate). Some useful graphical tools include scatter plots, stem and leaf diagrams, histograms, box plots, time series plots and conditional displays.
ME 202 (Fall 2011)
9/ 10
9 / 10
The Data Stage
Store the Data
Store the data in an efcient manner so that it can be accessed and used easily in the analysis stage. Store the data in a format that is compatible with the statistical software used in the analysis stage.
ME 202 (Fall 2011)
10/ 10
10 / 10