See how COGNOS+SPSS seamlessly butt bi+data Mining

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061601. Webp.jpg "/>

With the advent of the era of big data analysis, all-round data analysis capability has become an indispensable competitive power of enterprises today. The enterprise's full range of data analysis capabilities based on the level of analysis and the division of functional areas typically include: regular reports, ad hoc queries, multidimensional Analysis (also known as drillthrough or OLAP), early warning, statistical analysis, forecasting, predictive modeling (predictive predictive model), and optimization. In layman's words, the company's full range of data analysis capabilities can help companies analyze: ' What happened in the past ', ' What's happening ' and ' what's going to happen ', which is what we often call the past, the present and the future.

But in practice, the analysis of what happened in the past and what is happening now belongs to the functional category of the BI platform (such as Cognos Bi), and the BI platform can respond well to the ' past ' and ' Now ' what's happening ' and ' what's going to happen ' is a functional category of DM platform (data mining, such as SPSS), and the DM platform predicts future business by building predictive models to help companies answer questions about what might happen in the future.

The enterprise's full range of data analysis capabilities is shown in the following steps:

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061602. Webp.jpg "/>

(1) General report:

Regular reports are well known, and they are usually produced in a certain period of time, documenting the facts that have occurred over a period or a certain range. They are useful for understanding the state of the business, but are not able to make long-term decisions accordingly. Standard reports are primarily used to answer questions like ' What's happening ' and ' when '. Typical standard reports include monthly or quarterly financial reports.

(2) Ad hoc query:

Ad hoc queries often ' answer ' some common business questions by asking for a series of data (combinations). Ad hoc reports are primarily used to solve problems such as ' How much ', ' how often ' and ' where '. A custom report that records the daily sales of each product is an ad hoc report.

(3) Multidimensional Analysis (also known as drillthrough or OLAP technology):

OLAP technology can help you learn more about the details, which helps customers manipulate their data to find answers to questions such as ' How much ', ' What ' and ' where '. OLAP technology mainly solves problems such as ' where is the problem ' and ' How do I find the answer to the question '. For example, to sort the call behavior of different types of phone customers, to find out their call characteristics need to be applied to OLAP technology.

(4) Warning:

You can be alerted when the problem occurs and can be noticed in the future when similar situations occur. Alarms can be given in the form of e-mail, network channels, scorecards, or dashboards. The process of alerting requires confirmation of the trigger point of attention and what action is required once the alarm is taken. For example, the sales director will receive an alarm when the sales situation and sales target gap are large.

(5) Statistical analysis:

We can run a few more complex analyses. For example, variance analysis and regression analysis. We can make some assumptions based on the data and then use the data to build statistical analysis models to ' answer ' these assumptions. The problem solved by statistical analysis is mainly ' Why the behavior/event occurred ' and ' what kind of opportunity I lost '. For example, banks want to know what kind of people are more likely to refinance their homes, and then they use statistical analysis methods.

(6) Forecast:

It helps to build the right inventory, so that neither out-of-stock nor overstock. The main solution to the prediction is ' what will be the future trend ' and ' if such a trend continues. ' For example, retailers can predict the sales of specific products for a particular store in the future, based on their history, and this is a time series forecast.

(7) Predictive modeling (Predictive predictive model):

If you have 10 million customers who need to do a direct mail, who is most likely to respond? How to effectively clustering existing customers? Which customers are most likely to lose? Predictive models can answer these kinds of questions. Predictive models are primarily concerned with what is likely to happen in the future and the impact of different projections on the business. For example, a merchant can predict which product a customer might be more interested in, and which customers are more interested in a particular product.

(8) Optimization:

Optimization often brings innovation, which enables enterprises to maximize revenue (profit) under limited resources. The optimization emphasizes the way to make better use of various resources. For example, under certain resource conditions, how to arrange and maximize income profit is to optimize the problem that needs to be solved.

In the enterprise real data analysis application, the traditional BI platform and DM platform are relatively independent to complete their respective work, because the BI platform and the DM platform itself are independent platform tools, all support connected to various data sources and show the final analysis results. Although the end user in the BI platform to see the results of the report very much want to be able to directly use a data mining model for related predictive analysis, but in the actual scenario, the user has to import the source data of the report into the data mining platform to build the relevant model, generate the forecast results and write back to the database, Connecting to the result table via the BI platform is presented on the front-end report, which creates a tremendous amount of constant and additional development effort for the end user.

If the enterprise uses COGNOS+SPSS's BI+DM platform solution, the situation will become much simpler. In response to the need for online data prediction after the business user browses the report, the user only needs to introduce the Cognos report data previously viewed as input in the SPSS modeling interface to carry out the following modeling process. After the model is created and run, the predictive analysis results can be directly output to the Cognos report, and for the subsequent prediction results of browsing, the entire process users do not need to manually manage the data connection, the system through the underlying interface has been achieved from the Cognos platform to the SPSS platform seamless flow. The overall operation process is as follows:

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061603. Webp.jpg "/>

The following will assume a scenario for a customer churn forecast, as follows:

(1) Create a source report for training models on the Cognos platform and a target report for predicting churn, both source and target reports will be entered as data sources for the SPSS data stream.

Example: The following is an example of a Cognos report (source report) with customer churn results and detailed information.

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061604. Webp.jpg "/>

(2) Create a stream in SPSS Modeler and select IBM Cognos BI as input from the source. Edit the IBM Cognos bi source parameter, select the address and report path to connect to the Cognos server, and after the setup is complete, you can click the ' Preview ' button in the top left corner to preview the contents of the report.

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061605. Webp.jpg "/>

(3) After the process of creating the flow in SPSS Modeler based on the IBM Cognos BI data source, run the build forecast model. This example uses a decision tree model.

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061606. Webp.jpg "/>

The following is a detailed description of the forecast model

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061607. Webp.jpg "/>

(4) Set the Cognos target report for prediction to the input of the predictive data stream, set the appropriate Cognos connection parameters in the IBM Cognos BI source, use the decision tree predictive model that has been trained to do the data prediction, and write the results back to the Cognos BI report output. IBM Cognos BI Export is a built-in output feature in SPSS modeler that can output predictive results directly to Cognos reports.

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061608. Webp.jpg "/>

(5) After the completion of the above forecast model, return to the COGNOSBI platform to directly open the output forecast results report. The above red box indicates the content of the forecast result field generated by the forecast model, including the loss forecast and the confidence percentage.

650) this.width=650, "alt=" Cognos "class=" Img-thumbnail "src=" http://image.evget.com/images/article/2016/061609. Webp.jpg "/>

The above is the use of COGNOS+SPSS to achieve online report prediction analysis of the relevant process, if you are interested in the implementation of the entire process, you can click on the link below to download Cognos and SPSS trial attempt.

Cognos product trial download Link: http://bigdata.evget.com/product/200.html

SPSS product trial download Link: http://bigdata.evget.com/product/168.html

See how COGNOS+SPSS seamlessly butt bi+data Mining