This standard describes the composition, drawing and application of scatter diagrams. Scatter diagrams are used to determine whether there is a correlation between two variables. JB/T 3736.6-1994 Scatter diagrams, commonly used statistical tools in quality management JB/T3736.6-1994 Standard download decompression password: www.bzxz.net

Machinery Industry Standard of the People's Republic of China
JB/T 3736. 1~-3736. 8-- 94
Statistical tools commonly used in quality management
Published on December 9, 1994
Ministry of Machinery Industry of the People's Republic of China
Implementation on October 1, 1995
Mechanical Industry Standard of the People's Republic of China
Statistical tools commonly used in quality management
Scatter diagram
Subject content and scope of use
This standard describes the preparation, finalization and application of the scatter diagram. The scatter diagram is used to judge whether there is a correlation between two variables. The scatter diagram constitutes
JA/T 3736. 6 --94
代带日3756.6—84
Scatter plot is a simple graphical technique for studying the relationship between two sets of related data. It is composed of a series of points distributed in a rectangular coordinate system. These points represent several pairs of data of the variable. See Figure 1, x
Medical Scatter plot
3 Drawing step
Select the analysis object. It can be the relationship between the quality value and the patient, the relationship between the quality value and the measured characteristic. The relationship between the cause and the idle rate.
The collected data should have more than 30 pairs.
Record the observation points and their names, sampling methods, sampling period, measurement methods, fixed instruments, observation rocks, environmental conditions, etc. . Establish a rectangular coordinate system on the paper, and mark the data at the corresponding positions in the coordinate system. From the group and the group, the small and large values ​​are drawn and marked on the horizontal scale (X axis and vertical scale). The lengths of the two scales should be the same. If there is a reason that one quantity is the original or predicted value of the other, then the horizontal coordinate (time) represents the original or predicted value. e: When a point that is clearly different from the other data appears on the layout chart, the cause should be checked so that the statutory measurement or error is not required. 4 Analysis and use of the whole layout chart
4.1 The analysis of the unique state of the layout chart is shown in Table 1. Ministry of Machinery Industry 1994-12-09 Approved
1995-10-01 Approved
JB/T 3736.6-94
and the beauty system
one degree six, in order to study the big (strong positive constant correlation) change big, small (strong beauty)
An big, roughly change big (strong positive correlation) big change small (negative send
and: in what system (unexpected correlation)
and the people is the beauty rate "curve I related) Bin
change big·also corresponding to the exchange person·and the number of inserted signs, fast sex family better (strong stop correlation) Shi time degree again small teaching forbidden state fee
The continuity is good (strong correlation)
This standard stipulates that there is no constant correlation in the common method
The correlation effect does not need to be calculated
And the business model is used to make a judgment again,
The logarithm of the plot is extracted, and the integral value is returned to the national 4.2. When the conclusion is obtained by the pressure distribution, the original conditions remain unchanged, and the new verification is required. 31
Correlation test
JR/T 3736.6-94
When the points of the plot are weakly correlated or in a strong or weak correlation, a correlation test is required. .
Calculation of the relationship
The measured value range is in Table 2, and the values ​​in each table are calculated. The following changes will be calculated again
5.2 Correlation Analysis
The maximum value range of the correlation coefficient is
h. When>c, the average value also increases;
When the center increases and decreases;
d is small and small, when-=31,% and complete correlation:
f is larger, the better the correlation
, the distribution pattern of the method is different (see 4), 5.3 Sweep the data of Table S for testing.
When the variables are determined, there is a linear correlation at the level of the selected variable. At this time, the overall relevant items, this standard provides two and "risk balance for selection, the pressure of the 4 compensation is small, then the two concepts are required to be close to each other. The method of using the whole chart for loop communication and control is shown in B (special precautions) when the scattered chart should be used. See Appendix C for examples of reference), the table shows the periodic value of the correlation
0: 444
.. 258
0, 456
aota-50
-6.40:.n:
Method: r center. Become the bottom right rate tendency.
JB/T 373- 94
Layout type example
(examination material)
..2.....
.....:.
JB/T 3736.6-·94
Appendix B
Prediction and control methods using layout diagrams (reference material)
After determining that the two variables are linearly correlated, draw the product change: and, then draw a monitoring line. When the regression line and the actual situation are well fitted, the prediction or control of the change is in the range of the prediction or control.
BI draw a regression line on the scatter plot
Calculate the mean of 2.
Calculate the regression line for the whole series
And the regression line on the scatter plot. This line is limited to the cause of change of the original data and cannot be arbitrarily extended. See BI. x
B2 Draw the upper and lower control limits on the distribution of the standard deviation of the regression:
Calculate the control limits
Upper control limit
Lower control limit
JE/T3736.69
Method: 1! The control limits on the theory are shown: The standard limit is selected according to the number of monthly stops. The important considerations in the middle production are that when the price is falling and the amount is large, the
falls within the range of
, and the strong annual range is about 55.45%. The confidence within the range is 6826%
c. The upper and lower control limits are shown in the figure, see Figure B1. B3 Predict
for the given inference value range. See 2 pulse
D4 Control
When the requirement is to be in the heart. The change value should be controlled within the range shown in the figure or Figure 4. The control target cannot be achieved.
0, see Figure 3
If, then Figure B4.
Processing core
JB/F 3736. 6- 94
JB/T 3736. 6-94
Appendix
Application examples of the market chart
Reference materials!
Example: The quality of the gas produced by the generator is mainly determined by the content of carbon dioxide (CO), which is difficult to determine, while it is easier to determine the content of carbon dioxide. We hope that the content of CO can be known and finally the content of C can be measured. CI drawing
a: Analysis object. The relationship between the quality value C() and C content. b. Collect the 3C data of the production case or in Table C1. Set up a rectangular coordinate system and mark the data (,) in the corresponding positions, see Figure C1, Yt
Production date
Slow Yang language
April of ×
View
Co and co (the same point
×××
BC5.69:75.46
=7y, 81
-90, 4
Date of mourning
Analysis instrument
Analysis room
May 7, ×
Observation goods
Oil sample
800,89
1,6,11
Sheng Yankou porridge!
January, ×
Delayed position
Graphic analysis
JD/F3736.6-94www.bzxz.net
Continued table C1
coc's meaning system
765. 24 1 :3).75
From C1, we can see that the correlation coefficient between CO and CO, is calculated by connecting the items listed in Table 1.
b. Calculate L, L,, Lry
white date
-23818.29
Ln, = Zay
=—c. 93
. Calculate the relative coefficient
(195-32
135.3x845.3
V2.R1X0.53
-·0. 748
1.4 Correlation analysis
center, there is a negative correlation table between and,
analyzer
analyzer
global method
1274.21 238:8.29
Significance test of correlation coefficient
take=0.01-2=2
JB/T 3736. f.-94
table=453calculated result-0.1810463so there is almost a linear correlation between perfect 2 and precipitation.
C6Draw a regression line on the scatter plot
[calculate the work, the debt
and calculate the regression coefficient
28. =5 - (0. 333× 6. 51
List the regression straight line column
ga—tr
-- 3U. 34 -0. 33x
Draw a regression line on the scatter plot: see 2.
Tip: This standard content only shows part of the intercepted content of the complete standard. If you need the complete standard, please go to the top to download the complete standard document for free.