Application analysis of IBM SPSS Modeler Entity Analytics Example

Brief introduction

IBM SPSS Modeler Entity Analytics (EA) is a new feature added to IBM's SPSS Modeler 15.0 based on the IBM SPSS Modeler 14.2 Predictive analysis. Compared with traditional Modeler, Entity Analytics has a new dimension for data prediction. IBM's SPSS Modeler forecast focuses on predicting future behavior based on past data. The focus of Entity Analytics is to improve the coherence and consistency of current data by resolving its own identity conflicts in different existing data sources by parsing existing data.

function Introduction

EA Basic Configuration

If we have the following two data, data from a bank in the United States, and data from another bank B:

This third data, although it simply includes names, phone numbers, driver's licenses, and credit card numbers, is a powerful and effective way to link data A to data B. The names of data A and data C are all JOHN SMITH and the driver's license number is 0001133107;  The number of data B and C is 555-1212, and the credit card number is 102291328. So, we can infer that actually data A and data B actually describe the same person. If, in the future, we find that John Smith in data A is suspected of fraud, we can quickly infer that he may have acted the same way at Citibank and Morgan Bank. This early inference will have a very important effect on the prevention of the latter.

If there is only such a simple one or two-piece of data, we can see the correlation between the data at random eye view. But when you have tens of thousands of data, when your data comes from hundreds of different data sources, when your data is caught in malicious fake data, the human test is insignificant to no useful reason. and using our IBM SPSS Modeler 15 Entity Analytics new features, we use different input nodes to collect data from each data source together, and then through the Entity Analytics output contact all insert into our Entity an Alytics's exclusive database. Every time we have a new source of data, we compare it to all the known data in all databases by Entity Analytics's processing node, to determine whether they have a relationship or even infer how it is related. In addition, Entity Analytics can also help users to effectively merge information. For example, before my data, after the merger will have the following data:

In this way, we have a more comprehensive grasp of user Jon Smith's data, a comprehensive database will have a strong positive impact on the later marketing strategy.

Detailed information about the product can be obtained through the Help document, which is not mentioned here. Today, we would like to talk about its specific application case in business analysis.

Let's start by introducing our data sources, which include 850 rows, 28 columns, which we create ourselves based on the data component model of US credit card agencies. In this data, a very small number of people can be identified by traditional matching. In a model where the software is set to a traditional match, data that is judged to be the same person is automatically labeled as the same ea-id.

The simple Data preview screenshot is as follows:

Figure 1. Data Preview screenshot