The 66th Installment
A knack of data analysis
Hiroto Inoue,
Assistant Professor, Master Program of Innovation for Design and Engineering
Through my last studies, there were opportunities for undertaking the joint studies with peoples of various fields. The analyzed data range from result of subject evaluation experiment and trajectory of sight line (eye movement) using question papers to the financial statements of small and medium-sized enterprises. I would like to briefly explain about the knack of obtaining the significant information by data analysis.
The basic information provided by the result of data analysis:
- (1) It has a relationship / no relationship between something and something
- (2) There is a difference / no reference between something and something
For example, such information as “Products A and Products B are purchased at a time, and although Products C's sales volume is small than other products, the Products C is repeatedly purchased by the particular customers” can be used in the sales method as display pattern of the products and as adjustment of volume of the displayed products. The information is useful for a marketer. However, such useful information for the planner of new shop may not be the difference among the goods of his company but a differentiating factor when those are compared with the competitor’s. In the former case, the data for being analyzed is accumulated data in one’s company, but in the latter case, the competitor’s data is also required. In the circumstance of bothering analyzer, the first is ambiguity of purpose, “What you analyze it for?”, and it is the case that you are asked to analyze a data tentatively. To make hypothetical analysis of a subject or to make searching analysis of a subject, if there is no data for comparing, it is difficult to take a valuable conclusion.
In the circumstance of bothering the analyzer, the second is that he has little technical knowledge for collected data and does not gain the expert's cooperation. The technical knowledge does not mean the technique for data analysis but a phenomenon of subject of the analysis or property of data. For example, when data on debt are collected by question papers research, no response (missing value) arises or vanity is reflected in the response. Considering measures or preprocessing is necessary with estimation of the nature of data. When taste etc. is analyzed, if individual differences are distributed at random, the analyzer cannot predict. Individual differences itself may be important factor of analysis. If the analyzer would be an expert of the area, he collects necessary data with a clear purpose and chooses appropriate analysis method. If not, he may overlook a nature based data collection method without expert’s consultation. Team activity is important for data analysis and the request aforesaid “Will you please analyze this data tentatively” is such a problem.
If you discuss “It would be useful if we understand what specifically?” or “How we collect/collected data?” in the team before you analyze, it will be easy to draw a valuable conclusion from the result of data analysis . When you ask analyzer to analyze data, you would deepen your consideration of the task with the analyzer, not throwing things at him. It is the knack of data analysis and a request from analyzers around you.