Level: Intermediate
Before performing data analysis, it is necessary to create a dataset for analysis by extracting the necessary data, changing and shaping the correspondence between variables, and converting, categorizing, and recategorizing variables. JMP provides calculation functions necessary for database operations such as extracting rows and columns and merging, and for variable conversion. Using these functions, a dataset for analysis can be easily created. However, the analyst must instruct the operation instructions such as setting the extraction target and converting variables, and when the operation instructions are complex, the possibility of not obtaining the intended results increases. For example, when setting the range for data extraction or categorizing using if statements, the more complex the "and" and "or" rules, the higher the possibility of not obtaining the desired dataset for analysis. Therefore, it is necessary to mechanically check whether the dataset for analysis matches what the analyst intended.
JMP's statistical methods can be used to check the quality of a data set for analysis. Finding the maximum or minimum value of a variable is the simplest method, but the "distribution of one variable" and the "relationship between two variables" are also powerful methods, and a contribution ratio of 1 is evidence for the "relationship between two variables."
In this presentation, we will report on a case study in which we verified the quality of a dataset for analysis of large-scale data.
Lecturer in the Department of Industrial Engineering, Faculty of Science and Technology, Tokyo University of Science. Principal Researcher in the Department of Chemical Systems Engineering, Graduate School, The University of Tokyo. His research specialty is statistical quality control. He mainly studies statistical analysis methods required for quality control, but also applies statistical quality control to fireproof buildings, fire phenomena, and medical care and nursing care. He uses JMP to model large-scale data, extracts the information hidden behind it, and provides feedback to the specific field of research.
