PROF. MACQUEEN ONLY

Course MAT510_Week 4: Business Statistics- Statistical Engineering: Tactics to Deploy Statistical Thinking

Slide #

Slide Title

Slide Narration

Slide 1

Introduction

Welcome to Business Statistics.

In this lecture we will discuss statistical engineering in greater detail and tactics to deploy statistical thinking.

Slide 2

Topics

The following topics will be covered in this lesson:

Statistical Engineering Concepts;

Tools and techniques used in case studies;

Basic process improvement framework;

Identify root cause and lessons in case studies;

Basic problem-solving framework; and

DMAIC framework

Slide 3

Statistical Engineering

Statistical engineering studies how to best utilize statistical concepts, methods and tools and integrate them with information technology to improve process results. We can use information technology to formally embed the solutions into normal business flow. For example, if we develop a statistical model to evaluate the risk in potential financial deal, we can use information technology to automatically populate the model outputs and parameters so that a business underwriter can see all necessary information in his approval process.

Slide 4

Case Study: Reducing Resin Output Variation

In this case study, a production team at Ricoh’s Numazu plant tried to solve a perplexing problem - that their resin yield, which is computed by the ratio of actual production to theoretical outputs, is greater than 1. This is technically impossible. The team believes it is caused by output variations. Therefore, the team initiated an output variation reduction project. The team mapped out the manufacturing process. The team used tools such as time series, histograms, cause and effect diagrams and scatter plots to identify and prioritize the causes. Corrective actions were taken to reduce the errors. Although the desired average number was not obtained, reduction in output average variation was achieved. The team used subject matter expert knowledge and proper graphics tools to capture the improvement process in their system so that long term improvements were achieved.

Slide 5

Case Study: Reducing Telephone Waiting Time at a Bank

In this case study, a large bank wants to reduce customer call waiting time. The team identified causes for the long customer waiting times which are: the operator is busy helping other customers, the receiving party does not respond to the call quickly so the operator may have to locate them or find someone to take a message, and the operator does not know where to find the appropriate person to answer the customer. Using tools such as the cause and effect diagram or Pareto chart, they identified the root causes and implemented process improvement actions. The team brought in additional help for the operator. All employees must post a note on their whereabouts if they are not at their desk. The team generated an employee responsibility table for the operator. The team verified the improvement by taking data. It showed the process accomplished overall reduction in long waiting time by 80% and that the problem of having only one operator was reduced by 90%.

Slide 6

Basic Process Improvement Framework

Some logical steps and tools are listed in this step for the basic process improvement framework. The first step is to document and understand the process involved. The next step is to gather data on key outputs, inputs and process variables. It is critical to differentiate between stable and unstable processes. Stable process implies a lack of special causes, in which case we need to study and understand common cause structure so that we can fundamentally change the process. Instability implies the presence of special causes, in which case we need to identify and eliminate the root causes. The relationship between the cause and effect may be simple such as in the telephone operator case or complex such as in the resin variation reduction case. As one develops an understanding of causal relationships, it becomes easier to determine what changes are needed to improve the process. Once the process improvement framework has been implemented, we need to verify the impact of changes by gathering additional data.

Slide 7

Basic Process Improvement Framework (Continue)

The basic process improvement framework incorporates some key elements of statistical thinking strategy. They include improving results by improving the process; using the synergy between subject matter knowledge and data; diagnosing and reducing variations; and using sequential approaches. The basic process improvement framework has been proven effective in improving stable processes in a wide variety of business applications.

Slide 8

Case Study: Resolving Customer Complaints of Baby Wipe Flushability

In this case, the super-wipe brand by Scott Paper company suddenly lost a key feature in the product. It no longer flushed down the toilet. The team of experts was assembled to resolve the problem. This is a special variation case. The team generated a Manufacturing process flowchart to understand the problem. The subject matter experts took and examined data to identify priorities in their investigations. The causes include new synthetic fibers, use of X-Pro, and uses of chemical surfactants and trace chemicals in the plant water supply. They ruled out improbable causes such as synthetic fiber and impractical causes such as X-Pro usages. The break through happened when the process had to be shut down due to mechanical problems. When the crews restarted the plant, the flushibility problem improved significantly. The team found out that X-Pro was not applied according to the vendor’s instructions. They took corrective actions and established test procedures to avoid the problem and to allow rapid detection of re-occurrence.

Slide 9

Cast Study: The Realized Revenue Fiasco

In this case, the marketing manager’s December net realized revenue fell well short of budget. It is critical that they identify the root source and take corrective actions so that the revenue short fall near the year-end fiasco can never happen again. The marketing department data shows that unusually high rebates in December were the primary cause of lower realized revenue. Further investigation shows that rebates were high in December and sometimes in June. In the previous two years, the marketing department used a peaking sales technique that generated temporary high sales which masked the revenue short fall problem. This year, they did not use the peaking in sales. They also found that distributors like to push rebate paper near the end of the year or midway through the fiscal year. The root cause is a predictable special cause called structure variation. To resolve the problem, the marketing manager has to replace the information system. They design the new electronic billing system so that the rebate process can be completed efficiently. Other key lessons include that managing to budget is often inconsistent with the concept of continuous improvements. As in this case, the budget can be a restraint that prevents people from achieving even better results. To understand finer details of business processes, it is helpful to talk with people who actually work in the process.

Slide 10

Basic Problem-Solving Framework

Process improvement develops deeper understanding of normal behavior of the process so that the process can be fundamentally improved. In contrast, problem-solving identifies and diagnoses the root cause of the abnormal behavior in the process so that the process can return to normal. The problem solving framework include “Is”-”is not” analysis by documenting where (when, what, or who) it is or where (when, what, or who) it is not. Ask “5 Whys” from symptoms to reveal the root cause. Choose most plausible cause and most likely solutions. The key point of this step is to align the whole team on a common path forward. A common mistake is to declare success after implementing the solution before obtaining evidence that the problem has actually been solved. Therefore, we need to measure results. Even if the problem was solved, we are still not finished. It is very important to standardize the solutions to prevent future reoccurrence.

Slide 11

DMAIC Framework

Interaction with DMAIC framework

Introduction: The DMAIC framework came out of the Lean Six Sigma initiative which is a statistically oriented business improvement initiative used by companies such as General Electric and Bank of America. DMAIC consists of five main phases.

Directions: Click on each letter to learn more about each phase of the framework.

Define (D) select a project and carefully define the problem.

Measure (M) select the appropriate outputs to be improved based on customer guidance. We must ensure that they are quantifiable and can be measured.

Analyze (A) the preliminary data to document the current performance and to begin to identify root causes of problems and their impacts.

Improve (I) Determine how to intervene in the process to significantly improve the key metrics. Several rounds of improvements may be necessary.

Control (C) Once the desired improvements have been made, put in place a formal system to ensure that improvements are sustained even after the project is finished.

Slide 12

DMAIC Framework (Continue)

The typical tools used in the DMAIC phases are described in this figure. In coming weeks, we will discuss the details of these tools such as graphs, regression analysis, design of experiment and control charts. These tool are sequenced and linked in a unique way to enable users to identify the appropriate tools to use in the appropriate phase of the project. You should also note that DMAIC is unique from other statistical engineering frameworks in that it emphasizes project definition (Define phase) and standardization (Control phase), making these full phases equal in importance to the improve phase.

Slide 13

Check your understanding

Directions: Choose the best answer to complete the following sentence from the list below, and then click the Submit button.

Question: In process improvement, there are three main causes of variations. They are (1) common cause variations, (2) special cause variations and (3) structure variations. In process improvement, which cause should be resolved first? ___________.

A. Common cause

B. Special cause

C. Structure cause

Incorrect Choice A Feedback: Sorry – need to resolve special cause variation first.

Correct Choice B Feedback: Special cause variation needs to be addressed first.

Incorrect Choice C Feedback: Sorry – need to resolve special cause variation first.

Slide 14

Summary

In this lecture, we learned that statistical engineering tools are needed to produce tangible process improvements. From examples in the lecture, we understand the basic process to solve for process problems. We need to understand the problem and gather data on the problem. Suggestions from subject matter experts are important on these steps. Once the data is collected, we need to differentiate the root cause of the process problems. We need to solve the special cause first, since it hinders the resolution of common cause problems in the process. We need to re-design the process if structure cause becomes the bottleneck of the process improvements. These process steps are summarized in the DMAIC framework.