Pattern hex
School of Six Sigma

Graphs & Charts

When it comes to see­ing how process­es change over time, the abil­i­ty to graph and make charts of data can be invalu­able, espe­cial­ly for con­tin­u­ous improve­ment prac­ti­tion­ers. Graphs and charts can even help empha­size dif­fer­ent attrib­ut­es of the same data. But how do we know if we’re using the best visu­als for our par­tic­u­lar data set? This course will help with that decision.

What Do You Need to Know for Our Six Sigma Graphs Course?

Before we dive into the top Lean and Six Sig­ma graphs for busi­ness, con­sid­er this hypo­thet­i­cal scenario.

Your man­ag­er wants to gain a bet­ter idea of how a plas­tic mold­ing process is per­form­ing and asks for your help. She’s par­tic­u­lar­ly inter­est­ed in how con­sis­tent the weights of these plas­tic parts are.

To help you get start­ed, she pro­vides you with weight data of 200 parts that were ran­dom­ly sam­pled and weighed over the last few days of pro­duc­tion. With so much raw data to sift through, it can be dif­fi­cult to come to any accu­rate con­clu­sion. In this case, you decide graph­ing the data is the best way to make sense of it.

And though there are a num­ber of graphs to choose from, you set­tle on cre­at­ing a stan­dard his­togram. From this graph, you notice that the data seems to form a bell shape, which means the sam­ple mean is rough­ly four pounds. 

From this one graph alone, you are in a much bet­ter posi­tion to explain how this par­tic­u­lar process is per­form­ing to your manager.

Graphs on ipad

What You’ll Learn with Graphs

In this video, you’ll learn about the most pop­u­lar types of graphs used by Lean and Six Sig­ma prac­ti­tion­ers, as well as how and when to use them.

Keep in mind — many graphs are designed to work with vari­able data while oth­ers are designed to work with attribute data. And some of these graphs work bet­ter with larg­er sam­ple sizes, while oth­ers per­form well with small­er sam­ple sizes.

Pareto Chart

A Pare­to chart is a type of bar graph that has been sort­ed in order of decreas­ing fre­quen­cy, and often depicts which sit­u­a­tions are more sig­nif­i­cant. This chart gets its name from Ital­ian econ­o­mist and social­ist, Vil­fre­do Pare­to, who observed in 1906 that approx­i­mate­ly 80% of the land in Italy was owned by 20% of the pop­u­la­tion. Busi­ness con­sul­tants such as Joseph Juran lat­er applied this Pare­to Prin­ci­ple to qual­i­ty con­trol by dis­cov­er­ing that 80% of all effects come from 20% of the causes.

Gen­er­al­ly, Pare­to charts are used to iden­ti­fy improve­ment oppor­tu­ni­ties, par­tic­u­lar­ly when any kind of defect infor­ma­tion is encountered.

Pie Charts

One of the most pop­u­lar graphs on this list, pie charts are intend­ed to dis­play the rel­a­tive fre­quen­cy of count data, or whole num­bers or inte­gers. In oth­er words, pie charts are best for com­par­ing the rela­tion­ship of parts to a whole.

When­ev­er you hear per­cent of…” or part of… ‚” that’s an indi­ca­tion that a pie chart could meet your needs.

Histogram

If you want to bet­ter under­stand the shape and spread of vari­able data, or data that is acquired through mea­sure­ments, his­tograms are the way to go. The hor­i­zon­tal, or X axis of a his­togram, rep­re­sents the sale of a data. This scale is then divid­ed into equal inter­vals called bins. These bins act as buck­ets that hold data with­in a spe­cif­ic range.

When­ev­er data is nor­mal­ly dis­trib­uted, his­tograms will gen­er­al­ly have a bell shape to them, like in the exam­ple men­tioned ear­li­er. The bell shape con­veys that the data is fair­ly sym­met­ric since each side is essen­tial­ly a mir­ror image of the oth­er side.

While it’s pos­si­ble to cre­ate his­tograms with small­er sam­ple sizes, they may not be a good rep­re­sen­ta­tion of the true process. A good best prac­tice is hav­ing 50 or more data points to ensure accuracy.

Dot Plot

As its name implies, dot plots are graph­i­cal dis­plays of data using dots. This type of graph is rec­om­mend­ed when deal­ing with small sam­ple sizes of vari­able data.

While the dot plot may not help us under­stand the over­all shape of the data, we can dis­cov­er whether our data seems to be clus­tered in a spe­cif­ic region or whether we have out­liers or unusu­al data that war­rants fur­ther investigation.

Individual Value Plot

Instead of sum­ma­riz­ing the results into box­es and whiskers like dot plots, indi­vid­ual val­ue plots rep­re­sent each val­ue of data indi­vid­u­al­ly as a dot.

With these types of graphs, our ver­ti­cal, or Y axis, rep­re­sents the scale of the data. Each data point is then plot­ted accordingly.

Box Plot

Often used in explorato­ry data analy­sis, box plots help us see larg­er pat­terns in the dis­tri­b­u­tion of vari­able data.

Although box plots may seem anti­quat­ed in com­par­i­son to a his­togram, they have the advan­tage of tak­ing up less space, which is use­ful when com­par­ing dis­tri­b­u­tions between mul­ti­ple groups or datasets.

Time Series Plot

A time series plot allows you to dis­play data col­lect­ed in a time sequence from any process. This can help you track how your data is trend­ing over time and if the data points are exhibit­ing any patterns.

Control Chart

Con­trol charts, also known as She­whart charts or process behav­ior charts, are used to iden­ti­fy whether a man­u­fac­tur­ing or busi­ness process is in a state of control.

In this case, upper and low­er con­trol lim­its are added in order to help you under­stand the vari­a­tion in the process.