"Type 3" is the linked data among more than two type 2 data.
If we make good type 3 data, the best way to analyze is the graph. ( Graphs of Disribution and Line Graph )
Vlookup of Excel is useful to link data easily, if there are same row, "batch no", "lot no" and so on.
If there is not a same raw, time data is used. Knowledge of the background of the data ( meta knowledge ) is needed when we use time data.
Multi-Regression analysis and decision tree are useful for type 3.
But these methods do not deal with the information of time-series, the line plot is also needed.
By the analysis of type 3, we understand the fact we do not find in the analysis of type 1 .
If we analyze type 3 data again with the knowledge of the analysis of type 3 data, we may find what we did not notice on the first analysis.
The conclusion of the sensor data analysis goes perfect, if the conclusion is made by the total facts from type 1 analysis to type 3 analysis.