Tools Modified

Pareto Chart Provides the ability to

A type of bar chart to view causes of a problem in order of severity, largest to smallest o Categorize items, usually by content or cause factors o Identify the causes and characteristics that most contribute to a problem o Decide which problem to solve or basic causes of a problem to work on first o Understand the effectiveness of the improvement by doing pre- and postimprovement charts

Deployment

o o o o o o o

Define the problem clearly Collect data Sort or tally data in descending order Construct chart Draw bars to correspond to sorted data in descending order Determine vital few causes Compare and select major causes

Results

o o o

A necessary first step in continuous process improvement Graphically demonstrates the 20-80 Rule Provides the ability to identify which problem or cause to work on first by its severity or impact

Recommendations

It is easy to understand, but it requires discipline by management teams, facilitators, and teams involved

Pareto Voting

Deployment

Cause and Effect Diagrams Deployment

For problems that can’t be quantified, use Pareto Voting o Based on the Pareto Principle o 20% of the potential causes brainstormed will usually be chose by 80% of the group. Usually used with the cause and effect (fishbone) diagram o Complete brainstorming for potential causes of a problem. o Determine the total number of brainstormed ideas and multiply by 20%. For example: 10 ideas results in 2. If number is a fraction, round up to next whole number. o Based on the result of Step 2, determine the number of votes each team member receives. In this case, each team member receives two votes. o Each team member then uses his or her allocated votes (two in this case) to select the cause(s) having the largest impact on the stated problem. o Tally votes each cause receives. Those receiving the most votes are considered most important to the team. o Determine the plan of action to resolve these causes. Useful for visualizing, linking, identifying, classifying, and clarifying possible causes of a problem also known as fishbone diagram or characteristic diagram o Identify a problem (effect) with a list of potential causes o Construct a fishbone diagram with basic material – flipchart, paper, whiteboard, etc. o Write the effect at the right side o Identify major causes of the problems o Use Results of brainstorm or affinity diagram to fill in small branches o Complete process until lowest-level sub-cause is identified o Review, check, and verify with the work process that these causes do strongly affect the problem or causes being resolved

o o o

Select most important causes to work on first Can use many methods including the Pareto voting Verify root causes Continue this process to identify all validated root causes

Results o o o o

Check Sheets Deployment

Results

Advantage 

Provides a visual relationship between cause and effect Breaks down problem into manageable group of root causes that contribute most to a problem Separates symptoms of a problem from the real causes Provides interaction within a team to analyze problems

Check sheet is the technique or tool to record the number of occurrences over a specified interval of time o Clarify what must be collected o Establish the format for the data collection o Understand the objectives to ensure accuracy of the collection process o Establish the sample size and time frame of data collection o Instruct or train data collectors for consistency o Observe, record, and collect data o Tally results, using Pareto chart or histograms o Evaluate results o o

Provide factual data to evaluate problems Detects patterns in the process where the problems are suspect

Defines areas to discuss ,Limits scope , Consistency and Organized approach Over reliance ,Applicability and limiting

Dis Advantage - Histogram

Deployment

An orderly technique of grouping data by predetermined intervals to show the frequency of the data set Provide a way to analyze data collected Can display o Central point - Average o Variation - Standard deviation, range o Shape of distribution – normal, skewed, clustered o o o o  o o o o o

Gather data and organize from lowest to highest values Calculate the range (r): largest to the smallest Determine number of cells (k) Calculate the interval or width (m) of the cells m = r/k Sort the data or observations into their respective cells Count the data points of each cell to determine the height of the interval Create a frequency table Plot the results Distribution pattern from histograms: normal, double peak, isolated island, cliff, cogwheel, and skewed

o

Helps explain graphically if a process is in or out of control - define a process.

o o o o

Run Charts

Deployment

A run chart is a graph of data (observation) in chronological order displaying shifts or trends in the central tendency (average). The data represents measures, counts or percentages of outputs from a Process (products or services). A run chart is established for measuring or tracking events or observations in a time or sequence Order. o o o o o

Scatter Plot Diagrams

Deployment

Results

Provides a basis for what to work on first, especially using the Pareto chart application. Provides insight on the process capability to meet end user specifications. Establishes a technique to measure a process. Analyze for improvement opportunities.

Decide which outputs of a process you need to measure. Label your chart both vertically (quantity) and horizontally (time). Plot the individual measurements over time (once per time interval or as they become available). Connect data points for easy use and interpretation. Track data chronologically in time.

A scatter plot diagram shows the relationship that might exist between two variables or factors. It can test for possible cause and effect relationships. The use of this tool is for problem solving and the understanding of cause and effect relationships. It is often referred to as "correlation" diagrams (positive, negative, or zero). Typical scatter plot diagrams are: o Select the variable and response relationship to be examined by the team. o Gather data on variable and response; determine sample size of paired data. o Plot the results; determine appropriate scale to plot the relationship. o Circle repeated data points as many times as they occur. o o o

Gives analysis between two measurement variables in a process. Provides a test of two variables being changed to improve a process or solve a problem. Helps to recover real causes, not symptoms, of a problem or process.

Control Charts

A statistical technique to assess, monitor and maintain the stability of a process. The objective is to monitor a continuous repeatable process and the process variation from specifications. Typically used when the process seems to be out of control

Deployment

Identify characteristics of process to monitor Select the appropriate type of control chart based on characteristics to monitor o Determine methods for sampling – use check sheets o Collect the sample data o Analyze and calculate sample statistics o Construct control chart based on statistics o Monitor process for common and special causes o Evaluate and analyze any observations outside the limits for causes related to the situation Investigate unusual patters when observations have multiple runs above or below the average o Objectively defines a process and variation o Establishes measures on a process o Improves process analysis and opportunities o Process improvements are based on facts-managed by facts o o

Regression Analysis

Regression analysis is a means of showing the relationship between two variables. Regression analysis will provide two pieces of information. The first is a graphic showing the relationship between two variables. Second, it will show the correlation, or how closely related the two variables are.

Deployment

Regression analysis requires the use of a statistical software package, coupled with graphic presentation. The math required for regression analysis is usually substantial. Note that our regression analysis example is a regression of two variables. In actual practice, the analyst may wish to pick three or more variables in order to show regression. When more variables are used, it is important to calculate the standard error that is attributable to using the regression formula.

Results

o Cause and effect relationship of two variables, for example, the effect on cause of the size of a computer program. o Projecting performance based on multiple variables. For example, predicting what an individual's salary should be, based on the background, skill, and other performance characteristics of that individual. o Probing to determine the cause of some unfavorable event, such as computer projects significantly missing schedules. By developing correlations among multiple variables, the analyst can begin to see which variable may be causing the undesired result.