when we take on the data analysis, most of the time analysis, comparative analysis, analysis of these three kinds of segmentation methods in use, but there is actually a method we will often use cross analysis, especially in the investigation of abnormal data problems, cross analysis can show its strong power. Also want to say sorry to you is the blog of the frequency of updates may not be so frequent, but try every month at least to publish an article, I hope the quality assurance, welcome message or to initiate a discussion, some interesting topics, together with the expansion in the website data analysis method.
what is cross analysis?
cross analysis refers to the method of cross presentation of data in different dimensions and multi angle combination analysis. It can make up for some problems that can not be found by independent dimension analysis.
cross analysis with the multidimensional model and data cube based, can be regarded as a special kind of segmentation methods, but there is little difference with the breakdown of the concept, if interested can read papers before the data cube with OLAP. The method of subdivision is more in depth of the same dimension is based on drilling in OLAP (Drill-down), for example, every day from the month of data subdivision data is in the dimension of time segments, each city or province of view from the data in the data segment provinces, is the geographical dimension of the trip based on. The cross analysis is no longer confined to one dimension, like data cube and OLAP the cube, is based on the intersection of different dimension, time dimension, and the dimension of product performance in cross regional dimension analysis together each small cube data by OLAP slice and dice (Slice) (Dice) operation see for example in Shanghai City sales electronic products in March, it will help us to find many can not be found in a single dimension problem. Therefore, cross analysis is based on horizontal combinations of different dimensions, rather than subdividing them vertically in the same dimension.
The form of
cross analysis involves a combination of multi dimension, although the charts and tables can be displayed, but because the chart can express limited data, and is not easy to cross multiple dimensions shown in cross analysis is not too common, usually in the form. We usually see in the table is usually called a two-dimensional table, the first column generally place a dimension, such as date, list of all kinds of index header (in fact, all the indicators can also be considered as a special dimension, the two dimension index dimension) that is composed of the ranks of the most common two-dimensional table. The two-dimensional table can be extended to show a richer dimension:
, as shown above, is the typical layout of a multidimensional cross analysis based on tables. There are multiple dimensions placed hierarchically in the ranks. If we show only one indicator, then here >