Comparison of the hottest JMP and Minitab two simp

Comparison of JMP and Minitab (II): simple regression analysis

last time, I saw someone make a real PK of JMP and Minitab based on the application of "basic statistical analysis", and came to the conclusion that JMP is far better than Minitab in terms of statistical professionalism and overall ease of use. The author quite agrees with the author's realistic style, and here continues to follow this principle, and compares JMP and Minitab again from another commonly used statistical application of "simple regression analysis"

input the same two columns of data "X" and "Y" into the latest version of jmp7 and minitab15 respectively to get the linear regression equation, the scatter diagram with regression line, the regression test report and the prediction interval of regression line

comparison item 1: convenience of operation

the operation path of JMP, which is particularly suitable to be a testing instrument for controlling product quality on the production line, is: main menu analyze> fit y by X, fit line in the pop-up menu of the initial report, and its operation is not complicated, as well as the related options such as qualified curve fit and qualified curve indiv in the linear fit pop-up report. The resulting report is shown in Figure 1; The operation path of Minitab is: main menu st with the increase of nickel content at> regression> fitted line plot, select display confidence interval and display prediction interval in options, and the resulting report and graphics are integrated as shown in Figure 2. There is no obvious difference in the time of operation, but the JMP operation mode makes people realize the progressive relationship between the operation steps, which is highly logical, while the Minitab operation is purely a group of relatively independent mechanical actions connected by the user with memory

Figure 1 JMP output results

comparison item 2: the overall effect of the output report

jmp naturally integrates the statistical analysis results and relevant graphics, and users can see it at a glance. The statistical analysis results of Minitab are displayed in the session window, and the relevant graphics are displayed in another independent graph window, which adds a bit of trouble to checking. If the data, content and times of analysis are more than one, this kind of trouble will be more unbearable

comparison item 3: specific content of statistical analysis

the output results of JMP and Minitab are consistent regardless of the coefficient of the regression equation, R2, significance test p value, etc., which shows that the statistical principles behind the two software are actually the same. If you observe more carefully, you will find that the decimal digits in JMP are reserved more than Minitab, and can be customized, which is more accurate and professional

Figure 2 output result of Minitab

comparison item 4: effect of statistical graphics

in the early stage of regression analysis, only the most basic scatter diagram needs to be observed. The graphic effects of JMP and Minitab are similar. However, in the prediction application stage of regression model, the display of confidence interval is very important. JMP can deepen users' understanding of prediction model by "interval shadowing". In contrast, Minitab is dwarfed

if we want to compare the effect of the marginal graph, the gap between the two is even greater. JMP only needs to select histogram borders on the basis of the original report, and the results are shown in Figure 3. It not only retains the characteristics of the original prediction interval, but also realizes the dynamic link between the scatter diagram and the histogram. Minitab needs to re select graph> MA from the main menu, press the ABS key of the digital display scale, and then press the Rgnal plot, and then complete it in a new graph window. The results are shown in Figure 4. Unfortunately, the characteristics of the original prediction interval have disappeared, and the effect of dynamic links between graphs has never been reflected

Figure 3 marginal graph of JMP

Figure 4 marginal graph of Minitab

comparison item 5: expansibility of statistical analysis

even for the content involved by both sides, we can also explore the depth involved to observe the difference between the two. Taking polynomial regression as an example, JMP can support up to sixfold terms, while Minitab is only cubic terms. Taking saving data as an example, JMP can save not only residual values and predicted values, but also prediction formulas. Minitab does not have the function of saving formulas. And so on. The only thing that can make Minitab save some face is that it will perform residual analysis a little faster than JMP

summarizing the results of the above five comparisons, all those who really understand regression will come to a consistent conclusion: JMP is far better than Minitab in "simple regression analysis". The correctness of this conclusion may not be well understood when we do some simple work, but with the in-depth analysis of the problem, this feeling will be more and more strongly felt

similarly, the author is willing to introduce jade with this article, hoping that more lovers who really understand statistics, need statistics for quality management, and Six Sigma projects can exchange views and improve together