Experience of OLAP application in data analysis

The basis of rational decision is data, and OLAP is an important and necessary tool in data analysis. Combined with their own work characteristics and experience, flops:

1. When to use multidimensional degree of OLAP?

Multidimensional OLAP, while making the data thorough, can also cause the risk of data fragmentation. Therefore, the use of OLAP must be used for purposeful use, such as in the validation or reversal of the hypothesis test when the comparison (control group) of the selection, or to explore a particular dimension of the target value. OLAP can increase the resolution of data, but it also makes data fragmented. In the data greedy psychology, I remember in the clear target and then use multidimensional OLAP.

In the absence of a clear purpose, but to explore the data for OLAP, we must first in the current computing performance allows, set the resolution of the data (such as layer dimension). Using OLAP to explore data.

Most of the time OLAP is for the entire team to help query, this part of the function should not be overly extended, because http://www.aliyun.com/zixun/aggregation/14206.html "" "" "The cost of computing is difficult to make the team members fully understand. Word: OLAP is used either explicitly or explicitly.

2.OLAP Purpose:

The question that comes with it is what is the clear purpose of OLAP? If the purpose of data analysis is to provide a logical basis for judging decisions, OLAP is the raw material that provides analysis for these logical bases. A sudden significant change in the value of a metric, our goal is not to say that all dimensions, all levels of indicator performance, this is a inefficient method. Instead, it is advisable to overturn or support the optional hypothesis by using data in a highly correlated dimension that may affect the level. such as a index in the past 30 days in the chain fluctuation within 3%, but the last 2 days of fluctuations reached 10%, the exception to the first assumption, and then use the more granular data presented by OLAP to demonstrate or overthrow the hypothesis.

OLAP is hard to solve, but it is an important tool on the road to solving problems.

The above humble opinion, for reference only.