As stated in ISO 10017: Guide to Statistical Techniques for ISO 9001: 2000, the purpose of this document is to help organizations identify techniques for data analysis that may be useful in a Continuous Improvement process. and in solving the various problems they face.
In most processes the greatest enemy is variability, which can be observed in the quantifiable characteristics of products and processes, and exists in all stages of the product life cycle, the purpose of any organization is its control..
Statistical techniques, such as histogram, correlation analysis, etc., can help measure, describe, analyze, interpret, and model variability, even with a relatively limited amount of data. Statistical analysis of such data can help provide a better understanding of the nature, extent, and causes of variability. This could help solve and even prevent problems that may result from such variability.
continuous-improvement-toolThe techniques presented here can allow better use of available data to aid in decision making, and therefore improve the quality of products and processes to achieve customer satisfaction. These techniques are applicable for a wide spectrum of activities.
Our purpose is to help organizations learn about some of the appropriate statistical techniques for data analysis and problem solving.
For each of the selected techniques we include a definition of what " what is " technical, " to be used " and what are its " main benefit ". We also point out some of the " limitations " as well as " application examples " as well as propose a " way of developing" each tool and some " tips " on its use. At no time do we consider this document as unique or exhaustive, rather we invite the reader to look for other sources of information, such as the ISO 10017 standard itself and to seek professional help in the application and use of the techniques presented here. Grupo Kaizen SA offers courses related to these topics.
We ask readers to collaborate with examples, benefits, limitations and other topics that can help enrich the material that we offer to the business community free of charge, as a benefit of Grupo Kaizen SA
Instruction for the application of Cause Effect Analysis
- What is it:
It is a graphic representation in the form of a fish bone that allows us to identify the causes that affect a certain problem in a qualitative way. The cause-effect diagram is also known as the fishbone diagram or the Iskikawa diagram in homage to the name of its creator.
- For what do you use it:
It tries to discover in a systematic way the relation of causes and effects that affect a certain problem.
Additionally, it allows separating the causes into different branches or main causes known as the 4 M: Method, Labor, Machinery, Materials. In some cases other M's are included, such as the Middle and the Controls, but as originally proposed, the 4 are more than enough.
- Benefits:
The greatest benefit is that it allows to systematically focus on the causes that are affecting a problem and in a clear way to establish the interrelationships between those causes and the problem under study, as well as subdivide the main causes into primary, secondary and tertiary causes.
- Limitations and precautions:
Much depends on the prior knowledge of the people involved in the analysis. Sometimes it is also difficult to place a certain cause, which is not important.
The relationship is subjective, so it could not be said that these are really the causes of the problem.
- Application example: Causes of delay in deliveries
Product defects
Errors in the provision of services
Production Problems
- As elaborated:
A problem is selected and noted on the right side of a sheet of paper, enclosing it in a box. (Writing it down on the right side is only done by following the guidelines of its creator Kaouru Iskikawa, of Japanese origin, who as we know they write from right to left).
Subsequently, a horizontal line is drawn to the left of the box where the problem was enclosed, which is like the spine of a fish skeleton. Next, write the primary causes that affect the problem, in the form of large spines or lines and they are enclosed in a square.
The primary causes are: Materials (Raw Material, Information, documents), Machinery (Equipment, Software), Method (Procedures, instructions), Labor (Personnel, headquarters).
The secondary causes that affect the primary causes are written, followed by the tertiary ones.
Sometimes the importance is assigned to each factor and those that are particularly important appear to have a significant effect on the problem, according to the opinion of those involved in the problem. It is recommended to verify the relationship through data collection with the Inspection Sheet that we will see later.
7 Tips
Try not to go beyond what the group has control to avoid possible frustrations. If ideas are slow to come, use root causes to help: What is causing the (material) documents?
Be concise, use few words.
Make sure everyone agrees on the phrase that best describes the problem.
Ask yourself: Why does this cause happen?
Label with the necessary information, such as date, made by, process, etc.
CAUSE-EFFECT DIAGRAM
Instruction for the application of the Inspection Sheet
- What is it:
It is a tool that is used to collect the data of the problem that is analyzed. By designing a simple format, information is collected on indicators, causes of problems, etc. It is also known as a Verification Sheet or Check Sheet.
The inspection sheet in an information record indicating the number of times something has happened, for example the number of people served per hour at the till, the response time of promoters, the causes of returned checks, the cause of rejected requests, defects in products, etc.
The format must contain the following information:
- Service, department to which the data refers Date of collection and time if necessary
It is very important to determine the use to which the information will be put in order to establish the characteristics of the data and the collection format.
- For what do you use it:
In this type of format it is used to know the frequency with which the possible causes of the problems appear or also the frequency with which the clients appear during a certain period, as well as to record the time it takes to serve a client or a request. It can also be used to collect product weights, oven temperatures, etc.
- Benefits:
If it is well structured it allows you to collect information in a simple and practical way so that it does not interrupt the work of the person who is registering the information.
Lets answer the question When does it happen? Where does it happen? What does it consist of? Why is it happening? How does it happen? How often ?, as well as the origin of the data (Type of product, Process, box, department and person who took the data).
Facilitates tabulation of information.
- Limitations and precautions:
Care should be taken to record information in real time, which can be a problem if the inspection sheet is not well designed.
Application example:
Causes of delay in deliveries
Product defects
Errors in the provision of services
Errors in the preparation of checks, typing errors
Processing times at ATMs
Frequency of customer arrival in person or by phone
Furnace temperature
Products weight
- As elaborated:
The verification or inspection sheet can be as complicated or simple as the needs of the person who will use the information.
Determine the type of information you need to collect.
Establish the right amount of information to answer your questions.
Use the information you already have, whenever possible
Establish a unique data collection methodology and how to summarize it Pilot test and adjust the methodology if necessary
- Tips
Make sure that the data collection process is efficient, so that people have time to do it.
The data to be taken must be homogeneous (the same box, the same product, the same shift, the same machine,), otherwise, you will need to stratify the data (grouping).
Use the information from the Cause - Effect Diagram to make the Inspection Sheet.
Do not forget to complete with all the necessary information (Date, Department, Process, Person, Etc..)
INSPECTION SHEET
Instruction for the application of Pareto Analysis
- What is it:
It is a way of identifying and differentiating the few "vital", of the many "important" or giving priority to a series of causes or factors that affect a certain problem, which allows, through a graphic or tabular representation, to identify in a The aspects that appear more frequently or that have a higher incidence or weight are decreasing.
It can also be presented in other types of formats such as a "pastel" graphic.
Pareto analysis is also known as Law 20-80 which says that “generally a few causes (20%) generate the greatest number of problems (80%). It is also known as the ABC Law used for inventory analysis.
Its origin is due to studies carried out on people's income, by the Italian economist Wilfredo Pareto in the Middle Ages.
- For what do you use it:
It is used to establish where the greatest efforts should be concentrated in the analysis of the causes of a problem. This requires data, many of which can be obtained through the use of an Inspection Sheet. Below you will find an example of application in the Vital Clients Selection.
- Benefits:
The biggest benefit is that it allows you to focus on the causes that are really affecting the problem or to identify where efforts should be concentrated, such as analysis of sales by customers, by products, etc.
- Limitations and precautions:
When voluminous amounts of information are used, it requires the use of a computer and the graphic representation requires greater skill for an adequate representation. A limitation is that the most frequent or most expensive events are not always the most important: a fatal accident requires more attention than 100 finger cuts.
- Application example: Causes of delay in deliveries
Product defects
Errors in the provision of services
Production Problems
ABC Inventory Analysis
Customer Analysis
Accident analysis.
- As elaborated:
The list of causes, products or customers is ordered in decreasing order (highest to lowest) according to the frequency with which each of the causes was presented or the volume of sales by customer or by product. It is important to do it in the same unit of measure when it comes to products or customers. The most convenient is in monetary value.
The individual percentage of each category is calculated, dividing the value of each by the total of the causes or products.
The accumulated percentage is calculated, adding in decreasing order the percentages of each of the items in accumulated form.
If it is being used for the analysis of sales by product or by customers, the following rule applies: those products that are within the accumulated value up to 80% are called A. The following products that go from 80,001% to 95 % are called B and the rest until completing 100% are called C. This is what is known as the ABC Law or the 20-80 law, since approximately 20% of the products under study generate 80% of the total of sales.
Draw the graph:
Using a bar graph, order the causes from highest to lowest, recording the causes on the horizontal axis (X) and the values or frequency with which a certain cause occurred on the left vertical axis (Y). The percentage is noted on the right vertical axis. Excel allows you to make this type of compound chart.
7) Tips
Fill in the chart with the necessary information, such as the date, process analyzed, people who collected the information, etc.
Use common sense - the most frequent or most expensive events are not always the most important: a fatal accident requires more attention than 100 finger cuts.
Clearly identify the measurement standard ($,%, frequency). Use the data from the Inspection Sheet.
PARETO CHART
CAUSES
Instruction for the application of Pareto Analysis in the Selection of Vital Clients
- What is it:
It is a technique that allows classifying the vital elements for the company or for a department, be they clients, products, suppliers, services, etc.
- For what do you use it:
It is used to make a classification, depending on the volume of transactions in terms of quantity, monetary value, etc. as well as the degree of criticality or importance.
- Benefits:
It allows, in an objective way free of subjectivity, to classify those elements (clients, products, suppliers, services) to which the company or department should pay more attention either to establish a negotiation or to carry out a satisfaction survey. This scheme allows to make a higher classification than the one known as ABC, which in combination with criticality can give us a new classification of customers: Platinum, gold and silver, or alpha, beta, or range products.
- Limitations and precautions:
It is easier to establish an amount in terms of number of transactions or monetary value, but not in terms of the degree of criticality or importance that a little subjectivity can create.
- Application example:
Classify customers to develop service surveys.
Client classification to assign a mixed portfolio to an account executive.
Classification of clients for card delivery: platinum, gold, silver.
Classification of products in inventory for better administration or to establish purchasing policies or inventory level.
Classification of financial products.
Classification of services provided.
Classification of internal clients to discriminate the negotiation.
- As elaborated:
- ABC Classification: Make an ABC classification by the volume of transactions. Use the instruction for Pareto analysis. Classification 123: From the ABC classification list, each one is identified by its degree of criticality (Importance for the company's image, impact on the activities of the department, problems that could be caused by the lack of the product, distance from a supplier, (1) if it is a single provider and there are no substitutes, (2) if it is unique but there are substitutes, (3) if it is not unique and there are many substitutes). See sample listing below. Set the category: the category for each client, product or supplier is established as follows: Platinum Clients: A1.A2.A3.B1, C1, Gold Clients: B2, B3, C2, Silver Client: C3 or Alpha products: A1.A2.A3.B1, C1, Beta products: B2, B3, C2, Range products: C3.
List of (Clients, products, services, suppliers)
| Name | ABC | Criticality | Category |
| one | two | 3 | |
| TO | Alpha | Alpha | Alpha |
| B | Alpha | Beta | Beta |
| C | Alpha | Beta | Spectrum |
A: High volume
B: Medium volume
C: Low volume
1: Very critical 2: Critical
3: Uncritical
Platinum Clients: A1.A2.A3.B1, C1
Gold Clients: B2, B3, C2
Silver Client: C3
Alpha Customers: A1.A2.A3.B1, C1
Beta Clients: B2, B3, C2
Client Range: C3
7) Tips
For a classification of external customers, the volume of sales or business carried out can be a good way to do the ABC classification, for products in inventory (raw materials, inputs or finished products) both the quantity demanded and its unit price must be considered in a way. that the products are brought to a common monetary unit. In the case of services, it can be either for the value in sales or for the value of contribution to profits.
The degree of criticality, in addition to the aforementioned elements, may be due to the level of importance for the company in the case of customers (Regardless of their purchase volume, it is of interest for the image: for example, if the company wants to deliver cards to preferred customers, it could The President of the Republic has very little volume of transactions, but it is very important for the institution to count him as its first client and to deliver the card.
If a department wants to determine a sample size to interview in addition to the classification of its clients, it can use the following tips:
Instructions for calculating the sample size:
- Universe size or population size: For example, if you are calculating a sample of residents in a city of 50,000, the universe will be 50,000. Maximum acceptable error: is the probabilistic accuracy to be achieved. It is the statistical accuracy you need to achieve and the level of error you are willing to accept. The range can be between 1% and 20%. Desired level of confidence is used to determine the desired level of certainty for the results. For example, the set reliability level can be 95% or 90%. The desirable level of confidence determines to what extent you need to be sure of the reliability of the results. Normally 95% (1 in 20 probability of error) or 90% (1 in 10% probability of error) is chosen. When all the values are established, the result obtained will be the number of cases necessary to have representativeness of the universe or population with the levels of possibility of error and confidence that it established.
Examples: Sample to consider according to the size of the population of employees in a company.
Population size (Universe): 100
Maximum acceptable error
| Confidence level | two% | 3% | 5% |
| 95% | 96 | 91 | 79 |
| 99% | 97 | 95 | 87 |
Population size (Universe): 90
Maximum acceptable error
| Confidence level | two% | 3% | 5% |
| 95% | 87 | 83 | 73 |
| 99% | 88 | 86 | 79 |
Population size (Universe): 75
Maximum acceptable error
| Confidence level | two% | 3% | 5% |
| 95% | 73 | 70 | 63 |
| 99% | 74 | 72 | 67 |
Instruction for the elaboration of the Frequency Distribution and Histogram
- What is it:
It is the way in which the distribution of the measurements made in a process is represented, such as the hours of operation, maintenance response time, cement resistance, tube thickness, diameters, etc.
Frequency distributions can be presented in the form of a horizontal or vertical bar graph, but the groups need to be similar or homogeneous. The purpose of the frequency distribution is to analyze the data and obtain information on the behavior of a certain process.
- For what do you use it:
Use it when you need to discover and display the distribution of data by bar graphing the number of units in each category.
A histogram takes measurement data, for example, temperature, weights, dimensions, etc., and displays their distribution. This is critical since we know that all repeated events will produce results, which vary over time. A histogram reveals the amount of variation inherent in a process. The histogram is the first element to know the variability of a process.
It is used to know the way in which the data of a process or group of products is distributed, ages of a population, etc.
It allows to demonstrate how the data obtained from a sample serve as a basis for deciding on the population.
Organizing a good number of data in a histogram allows us to understand the population objectively.
For this it is necessary to have data, which can be obtained through the use of an Inspection Sheet
3) Benefits:
It allows to graphically show, by means of a histogram, the capacity of a process to satisfy the specifications or requirements of the clients.
It facilitates the understanding of how a process behaves and when there are special causes of variation.
Generally as a histogram it indicates that the greatest number of units are in the center, and that approximately an equal number of units is distributed on both sides. Many samples taken at random from data under statistical control follow this modality.
Other data shows distributions with all the data “stacked” at points far from the center, this type of distribution is known as “skewed”. It is important to remember that we will find distributions that should be normal and are not; the same can happen in distributions that are known in advance to be biased. In addition to knowing the form of distribution, you can know the following:
- If the "spread" of the curve falls within the specifications. If not, what quantity falls outside of them (Variability) If the curve is centered in the right place. We can tell if most of the data falls on the high or low side. (Bias).
- Limitations and precautions:
When voluminous amounts of information are used, it requires the use of a computer and the graphic representation requires greater skill for an adequate representation. Not useful in ring-controlled variables (fluids, some temperatures, etc.)
- Application example:
Frequency of the number of people in a certain schedule of attention, Frequency in the response time of departments such as maintenance, purchasing, personnel, Frequency with which certain values such as cement strength, tube thickness, diameters, weights etc. are presented.
- As elaborated:
For the elaboration of a Histogram, we will delve a little more in the instructions than in the other tools seen; This is due to the confusion that is created when deciding on the number of classes (bars), necessary or the class limits themselves, etc.
- Count the amount of data in the series (n).Determine the range, R, of the data. The range is the difference between the largest and smallest Value in the data set. Divide the value of the range ® by a certain number of classes referred to as K. The table below is one that shows us for different amounts of data the recommended number of classes to use. Determine the interval H, or class width: Amplitude.
A suitable formula to do this is as follows:
H = R / K
Round H to a suitable number. Do not forget that this interval must be constant throughout the entire frequency distribution. Another way to calculate it is K = 1 + 3.3 log no Square root of n.
- Determine the class limits:
For easy determination of class limits take the smallest individual measurement of the data. Use this number or round up to a smaller number. This will be the lower point of the first class limit. Take this number and add the class interval. Consecutive add the class interval to the lower class limit until you get the correct number of classes, which contains all the numbers.
- Define the class mark or midpoint ((LS-LI) / 2) Calculate the absolute frequency (Number of times a value is presented) Calculate the relative frequency (percentage of each value with respect to the sum of the total) Calculate the cumulative frequency (The cumulative sum of the individual percentages)
Histogram
It is a set of bars that represent the groups in a graph. The vertical line indicates the amount of data each group contains. The horizontal line marks the boundaries of all groups. A histogram is the graphical representation of a frequency table. The histogram shows us a quick view of the distribution of the measured characteristic. The histogram is a very important diagnostic tool since it shows a panoramic view of the variation in the distribution of the data.
The histogram reveals how much a process varies.
Histogram types:
- General type (symmetrical or bell shape):
The histogram mean value is in the center of the data range. The frequency is higher in the center and gradually decreases towards the ends. The shape is symmetrical. It's the most frequent form. It is known as normal distribution or bell.
- Comb type (bi modal)
Every third class has a lower frequency. This form occurs when the number of information units included in the class varies from one to another or when there is a particular trend in the way the data is approximated.
- Type with positive bias (with negative bias)
Asymmetric shape. The histogram mean value is located to the right (left) of the center of the range. The frequency decreases more sharply to the right (left), but gradually to the left (right). This form occurs when the lower (upper) limit is controlled theoretically or by a specification value or when values lower (higher) than a certain value are presented.
- Type of cliff to the left (of cliff to the right)
Asymmetric shape. The histogram mean value is located to the extreme left (right) away from the center of the range. The frequency decreases sharply to the left (right), and gradually to the right (left). This is a form that occurs frequently when a 100% selection has been made due to low process capacity, and also when the positive (negative) bias becomes even more extreme.
- Plain type
Frequencies form a plain, because classes have more or less the same frequency except those at the extremes. This form is presented with a mixture of several distributions that have different mean values.
- Double peak type (bimodal)
The frequency is low near the center of the information range and there is a peak on each side.
- Beak type isolated
A small isolated peak is presented in addition to a general histogram. This is the form that occurs when a small amount of data from a different distribution is included, such as a process abnormality, measurement error, or inclusion of information from a different process.
7) Tips
Don't expect all distribution to be normal.
Analyze the type of distribution obtained and its location with respect to the allowed limits. Look at the distribution to know its variability
See if the distribution is bi modal (2 shifts, two machines, two processes) which would mean different data sources
Write down all the necessary information
The number of classes (bars on the graph) determines the type of image in the distribution.
The distributions of some processes are biased by nature Use the data from the Inspection Sheet
- FREQUENCY DISTRIBUTION
| GROUP | LOWER LIMIT | UPPER LIMIT | CLASS BRAND | ABSOLUTE FRECUENCY |
PERCENTAGE |
ACCUMULATED PERCENTAGE |
| one | ||||||
| two | ||||||
| 3 | ||||||
| 4 | ||||||
| 5 | ||||||
| 6 |
HISTOGRAM
F
RE
CU
AND
N
INC
(CLASS MARK)
Instruction for applying Correlation Analysis (Regression) or Scatter Diagram
- What is it:
It is a graphic representation that shows the relationship of one variable with respect to another (There is not necessarily a cause-effect relationship).
Regression analysis relates the performance of a characteristic of interest (usually called the “response or effect” variable also known as the dependent variable) with potential cause factors (usually called the “explanatory” variable, cause or independent variable ”). Such a relationship is specified by a model that can be from science, economics, or engineering, etc., or can be empirically derived. The objective is to help understand the potential causes of variation in response and explain how each factor contributes to that variation. This is achieved by statistical relationship of the variation in the dependent variable with a variation of the cause or independent variable and obtaining the best fit by minimizing the deviation between the predictive and the actual response.
- For what do you use it:
The scatter diagram is used to study the possible relationship between two variables. This type of diagram is used to test possible relationships between cause and effect; It does not allow to prove that one variable is the cause of the other, but it does clarify if there is a relationship and the intensity that it could have.
Regression analysis allows you to do the following:
Test hypotheses about the influence of a potential cause variable on the response, and use this information to describe the estimated change in response for a given change in the cause variable;
Predict the value of the dependent variable for specific values of the independent variable;
Predict (at an established level of confidence) the range of values within which the response is expected, given the specific values of the cause variable;
Estimate the direction and degree of association between the dependent variable and the cause, independent or explanatory variable (such association does not imply causality). The information can be used, for example to determine the effect of changing a factor such as the temperature of a process, while other factors remain constant.
When you want to establish relationships between two indicators (result indicators and performance inducers), as suggested by the Balanced Scorecard (BSC) methodology
- Benefits:
Regression analysis can provide the relationship between various factors and the response of interest, and such a relationship can help guide decisions related to the process under study and ultimately improve the process.
Regression analysis has the ability to describe behaviors in response to consistent data, compare different subgroups of related data, and analyze potential cause-and-effect relationships. When relationships are well designed, regression analysis can provide an estimate of the relative magnitudes of the effect of the independent variable or cause, as well as the relative strength of those variables. This information is potentially valuable in controlling or improving the output of a process. Regression analysis can also provide an estimate of the magnitudes and source of influence on the response that comes from factors that have not been well measured or omitted from the analysis. This information can be used to improve the measurement system or the process.
Regression analysis can be used to predict the value of the dependent variable, for given values of one or more independent variables; it can also be used to predict the effect of changes in cause variables on an existing or predictive effect. It may be useful to conduct such an analysis before investing more time or money in a problem, when the effectiveness of the action is not known. In the BSC, it allows evaluating whether the improvement in a performance inducer really has any relationship with the result indicator.
- Limitations and precautions:
When designing a model, special knowledge is required to specify the appropriate regression analysis (eg, linear, exponential, multivariate etc), and in diagnostics to improve the model. The presence of omitted variables, measurement errors, and other sources of unexplained variations in the response can complicate the model used. The assumptions behind the regression model used, and the characteristics of the available data, determine which technique is appropriate for the analysis of the problem.
A problem sometimes encountered in developing the regression models is the presence of data, the validity of which is questionable. Whenever possible, the validity of such data should be investigated since their inclusion or omission may influence the estimates of the model parameters, and therefore the response.
It is important to simplify the model by minimizing the explanatory or independent variables. The inclusion of unnecessary variables can overshadow the influence of explanatory or independent variables and reduce the precision of model prediction. However, omitting an important explanatory variable can seriously limit the model and usefulness of the results.
- Application example:
Regression analysis is used to model production characteristics such as outputs, quality performance, cycle time, probability of failure in tests or inspections, and various deficiency behaviors in the processes. Regression analysis is used to identify most of the important factors in these processes, and the magnitude and nature of their contribution to the variation in the characteristic of interest.
Regression analysis is used to predict the outputs of an experiment, or controlled prospective or retrospective studies, studies of variation in materials, or production conditions.
Regression analysis is used to verify the substitution of one measurement method for another, such as replacing a
Destructive testing or the time consuming method by a non-destructive or time saving.
Cause-effect relationship between indicators and inductors
Relationship between strategic objectives and process objectives.
Examples of nonlinear regression applications including modeling drug concentrations as corresponding time and weight functions; modeling of chemical reactions as a function of time, temperature and pressure, etc.
6) How it is made:
The scatter diagram is drawn so that the horizontal axis (x axis) represents the values of one variable and the vertical axis (y axis) represents the values of another.
Gather 50 to 100 pairs of data (X, Y) whose relationships you want to study. It is advisable to have at least 30 pairs of data. Build a table similar to the following:
| Data | Weight in kilos | Height in meters |
| one | 72 | 1.77 |
| two | 81 | 1.55 |
| 3 | 99 | 1.90 |
| x | X | x |
| 30 | 47 | 1.55 |
Draw a graph with the horizontal (X) and vertical (Y) axes of the same length and appropriate scales. The values should increase as you move up and to the right on each axis. The variable that is being investigated as a possible “cause” is generally located on the horizontal axis (x) and the variable identified as an effect on the vertical
Record the data pairs on the chart
If you notice the values repeating, circle that point as many times as necessary.
Identify if correlation exists and type
Complete with the name, date, author etc.-
If you want you can run the map by using Excell:
Once you have completed the two columns of data, make a graph using the XYB (Scatter), which shows the relationship between the two variables. o Go to any of the plotted points on the graph and right-click. o Select "Add trend line" o Select the type, generally "linear" is used. o Then go to “options” and check the last two options: “Present equation and present R value on the graph”. o The equation identifies the way to calculate a new value of Y o The value of “r” indicates the degree of correlation between the two variables.
7 Interpretation:
The graphed points form a certain pattern. The direction and union of the cluster gives you insight into the strength of the relationship between variable 1 and variable 2. The more this pattern resembles a straight line, the stronger the relationship between the variables. Those is logical since a straight line indicates that every time one variable changes, the other changes in the same way.
1.- Positive correlation:
An increase in "Y" depends on an increase in "X". If "X" is controlled, "Y" is naturally controlled, for example: training vs. performance. R = 0.9 2.- Possible positive correlation:
If "X" increases, "Y" will increase a little, positive although "Y" seems to have causes other than "X". R = 0.6 3.- Non-correlation:
There is no correlation. "Y" may depend on another variable. R = 0.0 4.- Possible negative correlation:
An increase in “X” will cause a negative tendency to decrease in “Y” for example quality vs. customer complaints, training vs. rejections. R = 0.6 5.- Negative Correlation:
An increase in "X" will cause a decrease in "Y" therefore as in point 1, "X" can be controlled instead of "Y". R = 0.9
8 Tips
A negative relationship (If "y" increases, "x" decreases), it is as important as a positive relationship (If "x" increases, "y" increases)
It can only be said that "y" and "x" are related, not that one is the cause of the other.
Statistical tests are available to test the exact degree of relationship between variables.
It is always a good idea to look at the graph.
Generally straight line correlation is used where y = a + bx. However, this is not the only type of relationship that is usually found: there are other relationships such as logarithmic, exponential, etc., y = e, y = x 2, y 2 = x
Instruction for the application of Control Charts
1) What is:
It is an indispensable tool for detecting problems, as it provides information on variability due to its own or non-process causes and allows determining if it is under control.
- Indicates changes in the process Shows the presence of special causes of variation
- For what do you use it:
It is used to record data of a certain process where you want to measure variables such as delivery time, number of transactions, and various values such as weights, dimensions, temperatures, etc.
Its most frequent use is to control processes or to present information collected over a period of time.
Differentiate when the process is affected by normal causes of variation or causes unrelated to it.
- Benefits:
It allows to visually detect the trends of a certain process or established goal, shows if goals or specifications have been met at both the upper and lower levels, and serves to compare with other departments or companies.
A deeper analysis, using statistical techniques, allows detecting possible changes in the processes.
Its greatest benefit is controlling the processes and determining when to take action or do nothing.
Avoid using end-of-line inspection by attacking the problem before it occurs.
- Limitations and precautions:
The data must be collected at the time it is presented trying to respect its sequence. It can be used for both variables and attributes but requires more knowledge on how to apply each type of graph.
The type of graph must be studied correctly and how to apply it since there are several types: individual value graph, simple average, moving average, range-average, standard deviation-average, percentage graph, defective parts, defects and defects by Unit.
- Application example:
An example of this are the times of attention in boxes, times in processing operations, response time of the Promoters, delivery time of couriers, etc.
Number of operations attended.
Number of defects
Temperature control, weight control, filling control, dimension control. Etc.
- As elaborated:
The Y axis is the vertical line of the graph, which must contain the scale of values to be recorded as time, quantity, temperature, weight, number of errors, etc.
The X axis is the horizontal line (Time, hours, days, months).
A marked point indicates the measurement or quantity observed in a given time.
The dots must be connected to facilitate their interpretation.
The time period and the unit of measurement must be clearly identified.
If the recorded value is the average of several observations, the graph of averages must be accompanied by a graph of ranges.
Draw the average or average averages with a solid line.
Draw the upper and lower limits with a dotted line, these limits can be the established goal or the value that represents the performance of the competition or another department.
Statistically, the limits of normal variation correspond to plus or minus three standard deviations, or by other means, which depends on the graph, the data and the knowledge of those who apply statistical process control. More information can be obtained on the different ways of calculating limits.
- Tips
Make sure that the data collection process is efficient, so that people have time to do it.
The data to be taken must be homogeneous (the same box, the same product, the same shift, the same machine, the same process).
Use the information on the Inspection Sheet.
Do not forget to complete with all the necessary information (Date, Department, Process, Person, Etc.).
The order of the data must be maintained at the time it is collected.
Complete with all the necessary information.
Find additional information on the correct way to calculate control limits based on the graph used and the data type.
Data sheet
There are another series of techniques that can help continuous improvement and problem solving, such as the Deployment of the Quality function (Quality House), the Analysis of Failure Mode and Effects, the concept of statistical probability, the technique Six Sigma, as well as the Statistical Process Control (SPC), which if you require more information you can request [email protected] or request information about the seminars that Grupo Kaizen SA offers on all the topics mentioned here.
BIBLIOGRAPHY
- Manual of Basic tools for data analysis, Goal Basic Tools for Continuous Improvement, Hitoshi Kume How to apply the Deming Method, Mary Walton ISO 10017 Guide to Statistical Techniques for ISO 9001: 2000.
