Prediction Experiment for Dynamics 365 Business Central

By for October 26, 2018

Report Abuse
Use prediction model to fine-tune business processes with the power of machine learning and artificial intelligence. For example, reduce outstanding receivables and fine-tune your collections strategy by predicting whether sales invoices will be paid on time.
Using Model ----------- - Open this experiment in Azure Machine Learning Studio. - Run it, and then deploy it as a web service. - In the web service dashboard, copy the API key. Example API key: EXAMPLE42mSIPm4xkBqFuE8/8IiFxpRQI777wekuHu6YIPLmfe7WCrQpqWqvyOr1n8reaA517DaE5888GbSg== - Choose REQUEST/RESPONSE, and then copy the Request URI. Example endpoint URI: https://ussouthcentral.services.azureml.net/workspaces/xyz/services/xyz/execute?api-version=2.0&details=true - Add the API key and the request URI in Dynamics 365 Business Central: 1. For **Late Payment Prediction**, search for **Late Payment Prediction Setup** Model ----- The Prediction Model for Microsoft Dynamics 365 Business Central lets you easily train, evaluate, visualize models for prediction purpose . This model uses the Execute R Script module to run the R scripts that performs all the tasks. The two input modules define the expected structure of the input datasets. The first module defines the structure of the dataset, which is flexible and can accept up to 25 features, and the second defines the parameters. ![Prediction Model][1] Input Data Schema ----------------- When you call the api, you need to pass several parameters: - **method (String)** – Required parameter. Specifies the Machine Learning procedure to be used. The model supports the following methods: 2. train (system will decide whether to use classification or regression based on your dataset) 3. trainclassification 4. trainregression 5. predict 6. evaluate 7. plotmodel Based on selected method you may need additional parameters, such as - The **train_percent (Numeric)** – Required for **train**, **trainclassification**, and **trainregression** methods. Specifies how to divide a dataset into training and validation sets. 80 means 80% of dataset will be used for training and 20% for validation of result. - The **model (String;base64)** - Required for **predict**, **evaluate**, and **plotmodel** methods. Contents serialized model, encoded with Base64. You can get model as result of run train, trainclassification or trainregression methods. - The **captions (String)** – Optional parameter used with the **plotmodel** method. Contains comma separated captions for features. If not passed Feature1..Feature25 will be used. - The **labels (String)** – Optional parameter used with the **plotmodel** method. Contains comma separated alternative captions for labels. If not passed actual values will be used. - The **dataset** - Required for **train**, **trainclassification**, **trainregression**, **evaluate**, **predict**, consists of: - **Feature1..25** – The features are the descriptive attributes (aka dimensions) that describe the single observation (record in dataset). It can be integer, decimal, Boolean, option, code or string. - **Label** – Required but should be empty for predict method. The label is what you're attempting to predict or forecast Output Data Schema ------------------ The output of the service: - The **model (String;base64)** – Result of execution of **train**, **trainclassification**, and **trainregression** methods. Contains serialized model, encoded with Base64. - The **quality (Numeric)** – Result of execution of **train**, **trainclassification**, **trainregression**, and **evaluate** methods. In current experiment we use Balanced Accuracy score as a measure of a model’s quality. - The **plot (application/pdf;base64)** – Result of execution of plotmodel method. Contains visualization of model in pdf format encoded with Base64. - The **Dataset** – Result of execution of **predict** method, consists of: - **Feature1..25** – same as input. - **Label** – the predicted value - **Confidence** – the probability that classification is correct. Designed for Dynamics 365 Business Central ------------------------------------------ This model is especially valuable for applications that extend Dynamics 365 Business Central, which is a popular business management solution for small and mid-sized organizations. Highly adaptable and rich with features, Business Central provides processes in finance, manufacturing, sales, shipping, project management, services, and more. Through extensions, companies can easily add functionality that is specialized for their industry or region. Learn more and start free: https://dynamics.microsoft.com/en-us/business-central/overview/ See also: - [The Late Payment Prediction Extension][2] - [Forecasting Model for Microsoft Dynamics 365 Business Central][3] [1]: https://docs.microsoft.com/en-us/dynamics365/business-central/media/ml-models/PredictionModel.png [2]: https://docs.microsoft.com/en-us/dynamics365/business-central/ui-extensions-late-payment-prediction [3]: https://gallery.azure.ai/Experiment/Forecasting-Model-for-Microsoft-Dynamics-365-Business-Central