Graphing using Minitab L. Goch – February 2011
A GENDA Why Graph Data? Under STAT Run Chart Pareto Chart Multi-Vari Chart Under GRAPH Scatterplot Histogram Boxplot Individual Value Plot Bar Chart Pie Chart 3D Scatterplot All Minitab Tutorial Worksheets are located in the folder ‘C:\Program Files\Minitab 16\English\Sample Data’
Source: Donald Wheeler: Understanding Variation W HY G RAPH THE D ATA ? Graphs help us understand the nature of variation Graphs make the nature of data more accessible to the human mind Graphs help display the context of the data Graphs should be the primary presentation tool in data analysis If you can’t show it graphically, you probably don’t have a good conclusion Graphs help separate the signal from the noise Graphical Analysis is also Called DATA MINING!
R ULES FOR E FFECTIVE D ATA C OLLECTION Team must follow sampling plan consistently Do a short Pilot Run to test your procedures Note changes in operating conditions that are not part of the normal or initial operating conditions Maintain monitors on gauges for key process inputs Record any events that are out of the ordinary Log data into database quickly Keep a log book
A VAILABLE G RAPH T OOLS
R UN C HART : R UN C HART : S TAT > QUALITY T OOLS > R UN C HART
R UN C HART : S TAT > QUALITY T OOLS > R UN C HART Tests for Process Stability by applying some statistical diagnostic tests to the series Radon.mtw Open worksheet Radon.mtw
R UN C HART
P ARETO C HART : P ARETO C HART : S TAT > QUALITY T OOLS > P ARETO C HART
P ARETO C HART : S TAT > Q UALITY T OOLS > P ARETO C HART Pareto Charts are an essential tool to help prioritize improvement targets Pareto’s allow us to focus on the 20% of the problems that cause 80% of the poor performance EXH_QC.MTW Open worksheet EXH_QC.MTW
P ARETO C HART DefectsCounts Missing Screws274 Missing Clips59 Defective Housing19 Leaky Gasket43 Scrap4 Unconnected Wire8 Missing Studs6 Incomplete Part10
S ECOND L EVEL P ARETOS By We can generate a second level Pareto using the By statement This breaks down the overall Pareto by time of day
S ECOND L EVEL P ARETO FlawsPeriod ScratchDay PeelDay SmudgeDay ScratchDay OtherDay OtherEvening PeelEvening ScratchEvening PeelNight ScratchNight SmudgeNight ScratchNight PeelNight OtherNight ScratchNight PeelNight ScratchNight SmudgeNight ScratchNight OtherNight ScratchNight PeelWeekend SmudgeWeekend OtherWeekend
M ULTI -V ARI C HART : M ULTI -V ARI C HART : S TAT > QUALITY T OOLS > M ULTI -V ARI C HART
M ULTI -V ARI C HART : S TAT > Q UALITY T OOLS > M ULTI -V ARI C HART Multi-vari charts are a way of presenting analysis of variance data in a graphical form. The chart displays the means at each factor level for every factor. Sinter.MTW Open worksheet Sinter.MTW
M ULTI -V ARI C HART
S CATTER P LOT : S CATTER P LOT : G RAPH > S CATTER P LOT
S CATTER P LOT : S TAT > S CATTER P LOT Scatterplots study the relationship between two variables Batteries.MTW Open worksheet Batteries.MTW
S CATTERPLOT
S CATTERPLOT – B Y A V ARIABLE
H ISTOGRAM : H ISTOGRAM : G RAPH > H ISTOGRAM
C REATING A H ISTOGRAM WITH A N ORMAL C URVE Graph > Histogram > With Fit Histograms examine the shape and spread of data Camshaft.MTW Open worksheet Camshaft.MTW
S MOOTHED (N ORMAL ) D ISTRIBUTION We can view the data as a smoothed distribution (red line), in this example using the “normal distribution” assumption. It provides an approximation of how the data might look if we were to collect an infinite number of data points. DOES THE DATA FIT THE CURVE??? If not, does another type of distribution fit the data?
S MOOTHED (S KEWED ) D ISTRIBUTION We can view the data as a smoothed distribution (red line), in this example using the “skewed distribution” assumption. It provides an approximation of how the data might look if we were to collect an infinite number of data points. DOES THE DATA FIT THE CURVE??? If not, look for groups that may explain the shape of the data?
C REATING A H ISTOGRAM WITH G ROUPS Graph > Histogram > With Outline and Groups Data for the 2 different suppliers is available. Camshaft.MTW Still using worksheet Camshaft.MTW
S MOOTHED (S KEWED ) D ISTRIBUTION
B OX P LOT : B OX P LOT : G RAPH > B OX P LOT
B OXPLOTS : G RAPH > B OXPLOT There is another method of looking at the data that may be easier to see differences in the distributions Boxplots show the spread and center of the data BE CAREFUL! MEDIANMEAN The center of the Boxplot is the MEDIAN, not the MEAN Carpet.MTW Open worksheet Carpet.MTW
75 th Percentile 50 th Percentile or Median 25 th Percentile NOTE: Outliers will be displayed as * B OXPLOTS We can also generate boxplots by a variable to look at the variation due to that variable 75% to 100% 0% to 25% Average
B OXPLOTS W / G ROUPS We can also generate boxplots by a variable to look at the variation due to that variable Data for 4 Experimental Carpet types is available. Carpet.MTW Still using worksheet Carpet.MTW
B OXPLOTS W / G ROUPS
I NDIVIDUAL V ALUE P LOT : I NDIVIDUAL V ALUE P LOT : G RAPH > I NDIVIDUAL V ALUE P LOT
I NDIVIDUAL V ALUE P LOT : G RAPH > I NDIVIDUAL V ALE P LOT Individual Value Plots also show the spread and center of the data Billiards.MTW Open worksheet Billiards.MTW
I NDIVIDUAL V ALUE P LOT We can also generate Individual Value Plots by a variable to look at the variation due to that variable Average
I NDIVIDUAL V ALUE P LOT W / G ROUPS We can also generate Individual Value Plots by a variable to look at the variation due to that variable Data for 2 Additives is available. Billiards.MTW Still using worksheet Billiards.MTW
I NDIVIDUAL V ALUE P LOT W / G ROUPS
B AR C HART : B AR C HART : G RAPH > B AR C HART
B AR C HART : G RAPH > B AR C HART Bar Charts can be created from: 1) Data that needs to be counted 2) Functions of data(e.g. avg, min, max) OR 3) a Table
B AR C HART : G RAPH > B AR C HART (C OUNTS OF U NIQUE V ALUES ) Use to chart counts of unique values, clustered by grouping variables. Exh_QC.MTW Open worksheet Exh_QC.MTW
B AR C HART : G RAPH > B AR C HART (C OUNTS OF U NIQUE V ALUES )
B AR C HART : G RAPH > B AR C HART ( A F UNCTION OF A V ARIABLE ) Use to chart counts of unique values, clustered by grouping variables. Exh_AOV.MTW Still using worksheet Exh_AOV.MTW
B AR C HART : G RAPH > B AR C HART ( A F UNCTION OF A V ARIABLE )
B AR C HART : G RAPH > B AR C HART (V ALUES FROM A T ABLE ) asdfa Tires.MTW Open worksheet Tires.MTW
B AR C HART : G RAPH > B AR C HART (V ALUES FROM A T ABLE ) We can easily switch the X-axis so that CauseB is plotted within Qtr.
B AR C HART : G RAPH > B AR C HART (V ALUES FROM A T ABLE ) We can easily stack the Causes B into one bar on the X-axis still plotted within Qtr.
B AR C HART : G RAPH > B AR C HART (V ALUES FROM A T ABLE )
P IE C HART : P IE C HART : G RAPH > P IE C HART
P IE C HART : G RAPH > P IE C HART Use to display the proportion of each data category relative to the whole data set. Tires.MTW Open worksheet Tires.MTW
P IE C HART : G RAPH > P IE C HART
3D S CATTER P LOT : 3D S CATTER P LOT : G RAPH > 3D S CATTER P LOT
3D S CATTER P LOT : G RAPH > 3D S CATTER P LOT Use to evaluate relationships between three variables at once by plotting data on three axes. Reheat.MTW Open worksheet Reheat.MTW
3D S CATTER P LOT : G RAPH > 3D S CATTER P LOT Us the 3D Graph Tools to Enlarge & Rotate Graph (Check Tools >Toolbars >3D Graph Tools).
C ONCENTRATION D IAGRAMS CANNOT BE CREATED IN MINITAB Concentration Diagrams provide a visual display of occurrences to identify trends Usually a pictorial representation (drawing) of the product is used as the basis Occurrences are marked on the drawing where they were noticed for all units reviewed Take a look at the following examples… A Concentration Diagram is a great tool to Investigate the nature of surface defects
LOOKING FOR PAINT DEFECTS Top View of a Cooktop x X = 1 defect x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x x
A NNOTATING G RAPHS : To Change Title: Double click on Title, Change Font or Text, Click ‘OK’. To Add Subtitle or Footnote: Left Click anywhere on Graph, Click Add, Select Option to be added. To Underline Legend Title: Double Click on Legend box, Left click on ‘Header Font’ tab, Check Underline. To add data labels: Right Click anywhere on graph, Left click on ‘Add’, Left click on ‘Data Labels’, Left click on ‘OK’. To add Groups to data: Double Click on any Data Point, Select Groups tab, Select column to group by To Delete Legend Box: Right click on Legend box, Left Click on ‘Delete’. To move the position of a Label: Right Click to select the label you want to move. You may have to Right Click more than once. Right Click, hold and drag the label to the new position. To Unslant X-axis Labels: Double click X-axis, select Alignment tab, enter 90 for text angle, Click on ‘OK’. To Add Jitter to Data Points: Double click any Data Point, select the Jitter tab, Check Add jitter to direction, Click on ‘OK’.
C ONCLUSIONS Results need to be Supported by data Not based on conjecture or intuition 1) Graphical & 2) Statistical format Shown in 1) Graphical & 2) Statistical format 3) Engineering standpoint Make sense from an 3) Engineering standpoint Good Conclusions Require Data and Hard Evidence!! Good Conclusions Require Data and Hard Evidence!!