Thanks for your feedback.
Want to tell us more? Send us an email to leave your feedback.
After you have created a predictive model and generated PMML to represent that model, the next step is to import the PMML file into your MicroStrategy project. The import function creates predictive metrics from the PMML file.
In addition to importing predictive models directly into your MicroStrategy project, you can also import predictive models into the MDX cubes you have integrated into MicroStrategy, as described in Importing predictive models for MDX cubes.
|•||You have created a dataset report, which is the first step in creating a predictive model. For information on this process, see Creating a dataset report.|
|•||You have created and trained a predictive model from the dataset report, using MicroStrategy or a third-party application. For information on this process, see Creating a predictive model.|
|•||You must have the Create Application Objects and Use Metric Editor privileges to access the Import Data Mining Model option. For information on viewing or changing your privileges, see the System Administration Guide.|
|1||In MicroStrategy Developer, from the Tools menu, select Import Data Mining Model. The Import Data Mining Model dialog box opens.|
|2||Select the PMML file to import by clicking … (the Browse button), and then click Open.|
|3||You can make the following changes for the predictive metrics:|
|•||Select the Predictive Metric Type: The columns in this table are automatically populated when a predictive model is selected. You can change the information in this table as follows:|
|—||Predictor Type: By default, predictive metrics for all outputs defined within the model selected above will be created and displayed in this column. Clear any outputs for which you do not want to create predictive metrics.|
|—||Name: Displays the default names provided for each predictive metric, which can be changed.|
|—||Aggregation function: Displays the default aggregation functions for the selected predictive metric types. You can change these functions to determine how the output of the model should be aggregated. For more details, see Aggregating predictive metrics.|
|4||Click OK to import the model or models.|
|5||Select the folder in which to save all the predictive metrics.|
The import feature reads the PMML automatically and creates the predictive metrics in the folder specified. During the importing process, several key model characteristics are determined from the PMML, including the following:
|•||The type of data mining algorithm: Based on this, the appropriate MicroStrategy data mining function is used in the expression of the metric.|
|•||The inputs to the data mining function: Matching the column alias of the predictive input metric is important. Since Data Mining applications use this name to identify each variable, you can use the name of each variable in the PMML to identify which MicroStrategy metric should be used. The import feature robustly handles all potential circumstances in this area as follows:|
|▫||Only one MicroStrategy metric with that name exists: That metric is used automatically as an input to the function.|
|▫||Multiple MicroStrategy metrics with that name exist: The user is prompted to select the right metric.|
|▫||No MicroStrategy metrics with that name exist: The user is prompted to select a metric from within the project, or the user can cancel the import and try again later.|
|•||The version of PMML: Since different versions of PMML have different specifications, the version is identified so models are processed properly.|
|•||PMML validation: The PMML is validated for consistency with the PMML standard and known vendor-specific extensions.|
|•||Model verification: The Import dialog box detects whether the PMML contains verification data and reports this information at the bottom of the dialog box. Verification data can be viewed and tested for expected results by accessing the Model Viewer from the predictive metric.|
Once the predictive metric is created, you can use it in reports and other objects. See Using the predictive metric for more information.
Thanks for taking the time to give us some feedback.
Why wasn't this helpful? (check all that apply)
Vielen Dank, das Sie sich die Zeit genommen haben, uns Ihr Feedback mitzuteilen.
Warum waren diese Informationen nicht hilfreich? (Zutreffendes bitte auswählen)
Merci de nous avoir accordé un peu de votre temps.
Pourquoi ne vous ont-elles pas été utiles ? (cochez tous les choix applicables)
Grazie per avere dedicato del tempo per fornirci un tuo commento.
Perché non è stato utile? (selezionare tutte le risposte adeguate)
Gracias por dedicar su tiempo a proporcionarnos sus comentarios.
¿Por qué no ha sido útil? (Marque todos los que correspondan)
Obrigado por aproveitar o tempo para nos dar alguns comentários.
Por que isso não foi útil? (marque todas as que se aplicam)
피드백을 위해 시간 내 주셔서 감사합니다.
도움이 되지 않은 이유는 무엇입니까? (해당되는 모든 사항 선택)