@frontlinesystems/rason-mcp-server 2026.5.0

package/examples/datamining/Regression/Ensembles/Boosting.json

@@ -0,0 +1,63 @@
+{
+  "modelName": "BoostingRegression",
+  "modelDescription": "Regression Boosting Example: This RASON model example generates a fitted model and POSTs the model to the RASON server. Then uses the fitted model to score the training partition. This example uses one outside data source, hald-small.txt which may be downloaded by clicking the RASON Examples Download icon (download cloud icon). To solve this model, you must first POST the model to the RASON Server by clicking the POST icon (upload cloud icon on the left). Afterwards click Choose Files (on the Properties pane on the right), browse to the location of the downloaded examples, and select hald-small.txt. Then click the down arrow next to the Play icon and click Solve. For more information on POSTing and solving example models, click Help - Defining Your Data Mining Model.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model, in the datasources section. Typically these bind to external data sources due to data volume.",
+    "myTrainSrc": {
+      "comment": "Binds the dataset contained within hald-small.txt to the myTrainSrc datasource.",
+      "type": "csv",
+      "connection": "hald-small.txt",
+      "direction": "import"
+    }
+  },
+  "datasets": {
+    "comment": "Use the datasets section to bind to an entry in the 'datasources' section",
+    "myTrainData": {
+      "comment": "Binds the myTrainSrc datasource to the myTrainData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myTrainSrc",
+      "targetCol": "Y"
+    }
+  },
+  "weakLearner": {
+    "comment": "This section is only required when a bagging or boosting estimator is specified in 'estimator', and is used to define the weak learner used in the ensemble method.",
+    "treeWeakLearner": {
+      "comment": "Creates the weaklearner, treeWeakLearner, which selects decision trees as the weak learner and sets one parameter, minNumRecordsInLeaves. This weak learner will be used in the boosting estimator below. See the Rason Data Mining Model Components chapter in the RASON Reference Guide, downloadable from the Help tab.",
+      "type": "regression",
+      "algorithm": "decisionTree",
+      "parameters": {
+        "minNumRecordsInLeaves": 2
+      }
+    }
+  },
+  "estimator": {
+    "comment": "Define an estimator, including the type, algorithm and parameters. Only 1 estimator/transformer may appear in a RASON model.",
+    "boostingEstimator": {
+      "comment": "Defines the boostingEstimator as a boosting ensemble method estimator which uses the weakLeaner above, treeWeakLeaner; various options are set. For a complete list of all algorithm options, see the RASON Data Mining Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "type": "regression",
+      "algorithm": "boosting",
+      "parameters": {
+        "stepSize": 0.4,
+        "numWeakLearners": 2
+      }
+    }
+  },
+  "actions": {
+    "comment": "Define the actions, evaluations, and data to be used by the estimator defined above.",
+    "boostingModel": {
+      "comment": "This action fits a model using the myTrainData dataset using the regression boosting ensemble method estimator created above, boostingEstimator. The fitted model is exported to a JSON file for later use.  Two evaluations are requested in the result, weakLearnerModels (which returns the model for all weak learners) and numWeakLearners (which returns the number of weak learners). For a list of all available evaluations for Regression - Common Parameters, see the RASON Reference Guide downloadable from the Help tab.",
+      "data": "myTrainData",
+      "estimator": "boostingEstimator",
+      "export": "json",
+      "action": "fit",
+      "evaluations": [ "weakLearnerModels", "numWeakLearners" ]
+    },
+    "trainScore": {
+      "comment": "This action uses the fitted model, boostingModel, to score the myTrainData dataset (the training partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values and multiple statistics using the 'metrics'' evaluation (which returns sse, ss, sst, mse, rmse, mad, aoc and r2). See Evaluations: Regression - Common Parameters for a list of all evaluations supported.",
+      "data": "myTrainData",
+      "fittedModel": "boostingModel",
+      "action": "predict",
+      "evaluations": [ "prediction", "metrics" ]
+    }
+  }
+}

package/examples/datamining/Regression/Ensembles/RandomTrees.json

@@ -0,0 +1,79 @@
+{
+  "modelName": "RandomTreesRegression",
+  "modelDescription": "Regression Random Trees: This RASON model example generates a fitted model and POSTs the model to the RASON server. Then uses the fitted model to score both the training and validation partitions. This example uses two outside data sources, hald-small-train.txt and hald-small-valid.txt, which may be downloaded by clicking the RASON Examples Download icon (download cloud icon). To solve this model, you must first POST the model to the RASON Server by clicking the POST icon (upload cloud icon on the left). Afterwards click Choose Files (on the Properties pane on the right), browse to the location of the downloaded examples, select both files and click Open. Then click the down arrow next to the Play icon and click Solve. For more information on POSTing and solving example models, click Help - Defining Your Data Mining Model.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model, in the datasources section. Typically these bind to external data sources due to data volume.",
+    "myTrainSrc": {
+      "comment": "Binds the dataset contained within hald-small-train.txt to the myTrainSrc datasource.",
+      "type": "csv",
+      "connection": "hald-small-train.txt",
+      "direction": "import"
+    },
+    "myValidSrc": {
+      "comment": "Binds the dataset contained within hald-small-valid.txt to the myValidSrc datasource.",
+      "type": "csv",
+      "connection": "hald-small-valid.txt",
+      "direction": "import"
+    }
+  },
+  "datasets": {
+    "comment": "Use the datasets section to bind to an entry in the 'datasources' section",
+    "myTrainData": {
+      "comment": "Binds the myTrainSrc datasource to the myTrainData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myTrainSrc",
+      "targetCol": "Y"
+    },
+    "myValidData": {
+      "comment": "Binds the myValidSrc datasource to the myValidData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myValidSrc",
+      "targetCol": "Y"
+    }
+  },
+  "preProcessor": {
+    "comment": "This (optional) section may be used for preliminary data preparation or to compute values of some properties, which are passed later in the model.",
+    "numLeafRecords": {
+      "comment": "numLeafRecords equals the Maximum of 1 or the number of rows in the dataset devided by 10. numLeafRecords is used to set the value for the minNumRecordsInLeaves in the estimator below.",
+      "formula": "INT(MAX(1, ROWS(myTrainData) / 10))"
+    }
+  },
+  "estimator": {
+    "comment": "Define an estimator, including the type, algorithm and parameters. Only 1 estimator/transformer may appear in a RASON model.",
+    "randomTreesEstimator": {
+      "comment": "Defines the randomTreesEstimator as a regression random trees ensemble method estimator. Note that there is no need for the weakLearner section when using a random trees estimator. Various options are set. Note that numLeafRecords (computed above) is used to set the value of the minNumRecordsInLeaves option. For a complete list of all algorithm options, see the RASON Data Mining Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "type": "regression",
+      "algorithm": "randomTrees",
+      "parameters": {
+        "numWeakLearners": 2,
+        "bootstrapSeed": 1,
+        "featureSelectionSeed": 2,
+        "minNumRecordsInLeaves": "numLeafRecords"
+      }
+    }
+  },
+  "actions": {
+    "comment": "Define the actions, evaluations, and data to be used by the estimator defined above.",
+    "randomTreesModel": {
+      "comment": "This action fits a model using the myTrainData dataset using the regression random trees ensemble estimator created above, randomTreesEstimator. The fitted model is exported to a JSON file for later use. Three evaluations are requested in the result, weakLearnerModels (which returns the model for all weak learners), numWeakLearners (which returns the number of weak learners) and featureImportance (which returns the variables that are included in the model along with their Importance value). For a list of all available evaluations for Classification - Random Trees Ensemble Methods, see the RASON Reference Guide downloadable from the Help tab.",
+      "data": "myTrainData",
+      "estimator": "randomTreesEstimator",
+      "export": "json",
+      "action": "fit",
+      "evaluations": [ "weakLearnerModels", "featureImportance", "numWeakLearners" ]
+    },
+    "trainScore": {
+      "comment": "This action uses the fitted model above, randomTreesModel, to score the myTrainData dataset (the training partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predictions and statistics such as SSE, SS, SST, MSE, RMSE, MAD and R2. See Evaluations:Regression - Common Parameters for a list of all evaluations supported.",
+      "data": "myTrainData",
+      "fittedModel": "randomTreesModel",
+      "action": "predict",
+      "evaluations": [ "prediction", "sse", "ss", "sst", "mse", "rmse", "mad", "r2" ]
+    },
+    "validScore": {
+      "comment": "This action uses the fitted model above, randomTreesModel, to score the myValidData dataset (the validation partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values and multiple statistics using the 'metrics'' evaluation (which returns sse, ss, sst, mse, rmse, mad, aoc and r2). See Evaluations: Regression - Common Parameters for a list of all evaluations supported.",
+      "data": "myValidData",
+      "fittedModel": "randomTreesModel",
+      "action": "predict",
+      "evaluations": [ "prediction", "metrics" ]
+    }
+  }
+}

package/examples/datamining/Regression/FindBestModel.json

@@ -0,0 +1,87 @@
+{
+  "comment": "Prediction: Find Best Model RASON example - Find Best Model uses methods similar to those in (expensive high-end) tools like DataRobot and RapidMiner, to automatically choose types of ML models and their parameters, validate and compare them according to criteria that you choose, and deliver the model that best fits your data!",
+  "modelName": "FindBestModelPredict",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model here. These datasources must be bound to a dataset in the next section.",
+    "myTrainSrc": {
+      "comment": "Imports the bh-class-train dataset.",
+      "type": "csv",
+      "connection": "bh-class-train.txt"
+    },
+    "myValidSrc": {
+      "comment": "Imports the bh-class-valid dataset.",
+      "type": "csv",
+      "connection": "bh-class-valid.txt"
+    }
+  },
+  "datasets": {
+    "comment": "Use the datasets section to bind to an entry in the 'datasources' section",
+    "myTrainData": {
+      "comment": "Binds the myTrainSrc datasource to the myTrainData dataset.",
+      "binding": "myTrainSrc",
+      "targetCol": "CATMEDV"
+    },
+    "myValidData": {
+      "comment": "Binds the myValidSrc datasource to the myValidData dataset.",
+      "binding": "myValidSrc",
+      "targetCol": "CATMEDV"
+    }
+  },
+  "estimator": {
+    "comment": "Define an estimator, including the type, algorithm and parameters.",
+    "fbmEstimator": {
+      "comment": "Defines the Find Best Model prediction estimator, fbmEstimator. Use the learners property (optional) to set options for the various regression weak learners, including the 3 ensemble methods. This example sets the k for the nearest neighbors prediction method, the fit intercept option for multiple linear regression and the weak learner for the boosting ensemble method. Omit the learners property to use each learner's option default settings. See Algorithm Parameters: Find Best Model for Classification & Regression for a list of all options specific to the Find Best Model feature. See Algorithm Parameters for each learner for options specific to a learner, i.e. for options related to k-nearest neighbors, see Algorithm Parameters : k-nearest neighbors regression.",
+      "type": "regression",
+      "algorithm": "findBestModel",
+      "learners": {
+        "linearRegression": {
+          "fitIntercept": false
+        },
+        "nearestNeighbors": {
+          "numNeighbors": 5
+        },
+        "boosting": {
+          "weakLearner": {
+            "type": "regression",
+            "algorithm": "nearestNeighbors",
+            "parameters": {
+              "numNeighbors": 10
+            }
+          }
+        }
+      }
+    }
+  },
+  "actions": {
+    "comment": "Define the actions, evaluations, and data to be used by the estimator defined above.",
+    "fbmModel": {
+      "comment": "This action fits a model using the myTrainData dataset using the find best model estimator created above, fbmEstimator. The Find Best Model 'fit' action has only one possible evaluation, 'messages'.  This evaluation reports the fitting log which is where any warnings/failures that occur during the fitting process will be reported.",
+      "trainData": "myTrainData",
+      "estimator": "fbmEstimator",
+      "action": "fit",
+      "evaluations": [ "messages" ]
+    },
+    "trainScore": {
+      "comment": "This action uses the fitted model above, fbmModel, to score the myTrainData dataset (the training partition). If 'bestLearnerMetric' parameter is present in the 'predict' action. Then the best learner, for the given 'predict' action, according to the 'bestLearnerMetric' will be returned in the results. See Evaluations: Find Best Model for a list of all evaluations supported by Find Best Model.",
+      "data": "myTrainData",
+      "fittedModel": "fbmModel",
+      "parameters": {
+        "bestLearnerMetric": "R2"
+      },
+      "action": "predict",
+      "evaluations": [ "modelPerformance", "bestLearner", "learnerForScoring", "prediction" ]
+    },
+    "validScore": {
+      "comment": "This action uses the fitted model above, fbmModel, to score, or predict the values for the myValidData dataset (the validation partition). If 'bestLearnerMetric' parameter is present in the 'predict' action. Then the best learner, for the given 'predict' action, according to the 'bestLearnerMetric' will be returned in the results. See Evaluations: Find Best Model for a list of all evaluations supported by Find Best Model.",
+      "data": "myValidData",
+      "fittedModel": "fbmModel",
+      "action": "predict",
+      "parameters": {
+        "bestLearnerMetric": "R2",
+        "useForScoring": true
+      },
+      "evaluations": [ "modelPerformance", "bestLearner", "learnerForScoring", "prediction" ]
+    }
+  }
+}

package/examples/datamining/Regression/LinearRegression.json

@@ -0,0 +1,125 @@
+{
+  "modelName": "LinearRegression",
+  "modelDescription": "Regression: Linear Model: This RASON model example generates a fitted model and POSTs the model to the RASON server. This example uses three outside data sources, hald-small-train.txt, hald-small-valid.txt and hald-small-score-single-rec.txt. All may be downloaded by clicking the RASON Examples Download icon (download cloud icon). To solve this model, you must first POST the model to the RASON Server by clicking the POST icon (upload cloud icon on the left). Afterwards click Choose Files (on the Properties pane on the right), browse to the location of the downloaded examples, and select all three files.  Then click the down arrow next to the Play icon and click Solve. For more information on POSTing and solving example models, click Help - Defining Your Data Science Model.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model in the datasources section. Typically these bind to external data sources due to data volume.",
+    "myTrainSrc": {
+      "comment": "Binds the dataset contained within hald-small-train.txt to the myTrainSrc datasource. This dataset represents the training partition.",
+      "type": "csv",
+      "connection": "hald-small-train.txt",
+      "direction": "import"
+    },
+    "myValidSrc": {
+      "comment": "Binds the dataset contained within hald-small-valid.txt to the myValidSrc datasource. This dataset represents the validation partition.",
+      "type": "csv",
+      "connection": "hald-small-valid.txt",
+      "direction": "import"
+    },
+    "myNewSrc": {
+      "comment": "Binds the dataset contained within hald-small-score-single-rec.txt to the myNewSrc datasource. This dataset represents the test partition or the new data to be scored.",
+      "type": "csv",
+      "connection": "hald-small-score-single-rec.txt",
+      "direction": "import"
+    },
+      "myExportSrc": {
+        "comment": "Binds to the detailed residuals output which will be exported.",
+        "type": "csv",
+        "connection": "Results/Export/influence-diagnostics.csv",
+        "direction": "export"
+      }
+    },
+    "datasets": {
+      "comment": "Datasets are bound to an entry in the 'datasources' section",
+      "myTrainData": {
+        "comment": "Binds the myTrainSrc datasource to the myTrainData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+        "binding": "myTrainSrc",
+        "targetCol": "Y"
+      },
+      "myValidData": {
+        "comment": "Binds the myValidSrc datasource to the myValidData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+        "binding": "myValidSrc",
+        "targetCol": "Y"
+      },
+      "myNewData": {
+        "comment": "Binds the myNewSrc datasource to the myNewData dataset.",
+        "binding": "myNewSrc"
+      }
+    },
+    "estimator": {
+      "comment": "Defines an estimator, including the type, algorithm and parameters. Only 1 estimator/transformer may appear in a RASON model.",
+      "mlrEstimator": {
+        "comment": "Defines the mlrEstimator as a linear regression estimator; the fitIntercept option is set to True. For a complete list of all algorithm options, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+        "type": "regression",
+        "algorithm": "linearRegression",
+        "parameters": {
+          "fitIntercept": true
+        }
+      }
+    },
+    "actions": {
+      "comment": "Define actions, evaluations, and data to be used by the estimator created above.",
+      "mlrModel": {
+        "comment": "Fits a model to the myTrainData dataset using the mlrEstimator estimator. The fitted model is automatically POSTed to the RASON Server.This fitted model will be used to score new data in the Scoring Using Linear Regression - Fitted Model: JSON example in Examples- Data Science - Scoring. In addition, the 'export' parameter exports the fitted model to a JSON file. For a complete list of all available evaluations, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+        "trainData": "myTrainData",
+        "estimator": "mlrEstimator",
+        "export": "json",
+        "action": "fit",
+        "evaluations": [
+          "fittedModelJson",
+          "regressionSummary",
+          "detailedResiduals",
+          {
+            "name": "influenceDiagnostics",
+            "binding": "myExportSrc"
+          },
+          "multicollinearityDiagnostics",
+          "trainingIntervals",
+          "anova",
+          "coefficients",
+          "detailedCoefficients",
+          {
+            "name": "varianceCovariance",
+            "dataFrame": false
+          },
+          "predictorScreeningInfo",
+          "entranceTolerance"
+        ]
+      },
+      "trainScore": {
+        "comment": "Uses the fitted model mlrModel to score the training dataset, myTrainData (the training partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values, the residuals, SSE, SS, etc.. For a complete list of all available evaluations, see the Regression - Common Parameters section in the RASON Reference Guide. (Click the Help tab to download.) ",
+        "data": "myTrainData",
+        "fittedModel": "mlrModel",
+        "action": "predict",
+        "evaluations": [ "prediction", "residuals", "newIntervals", "sse", "ss", "sst", "mse", "rmse", "mad", "r2", "aoc" ]
+      },
+      "validScore": {
+        "comment": "Uses the fitted model mlrModel, to score the validation dataset, myValidData (the validation partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values, the redisiduals, and various metrics (SSE, SS, SST, MSE, RMSE, MAD, R2 and AOC). For a complete list of all available evaluations, see the Regression - Common Parameters section in the RASON Reference Guide. (Click the Help tab to download.) ",
+        "data": "myValidData",
+        "fittedModel": "mlrModel",
+        "action": "predict",
+        "evaluations": [
+          {
+            "name": "prediction",
+            "dataFrame": false
+          },
+          "residuals",
+          "newIntervals",
+          "metrics"
+
+        ]
+      },
+      "newScore": {
+        "comment": "Uses the fitted model mlrModel to score the training dataset, myTrainData (the training partition) using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values, the redisiduals, etc.. For a complete list of all available evaluations, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.) ",
+        "data": "myNewData",
+        "fittedModel": "mlrModel",
+        "action": "predict",
+        "evaluations": [
+          {
+            "name": "prediction",
+            "dataFrame": false
+          }
+        ]
+      }
+    }
+  }

package/examples/datamining/Regression/NearestNeighbors.json

@@ -0,0 +1,78 @@
+{
+  "modelName": "NearestNeighborsRegression",
+  "modelDescription": "Regression: K Nearest Neighbors; This RASON model example generates a fitted model and POSTs the model to the RASON server. Then uses the fitted model to score both the training and validation partitions. This example uses two outside data sources, hald-small-train.txt and hald-small-valid.txt. Both may be downloaded by clicking the RASON Examples Download icon (download cloud icon). To solve this model, you must first POST the model to the RASON Server by clicking the POST icon (upload cloud icon on the left). Afterwards click Choose Files (on the Properties pane on the right), browse to the location of the downloaded examples, and select both files. Then click the down arrow next to the Play icon and click Solve. For more information on POSTing and solving example models, click Help - Defining Your Data Science Model.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model, in the datasources section. Typically these bind to external data sources due to data volume.",
+    "myTrainSrc": {
+      "comment": "Binds the dataset contained within hald-small-train.txt to the myTrainSrc datasource. This dataset represents the training partition.",
+      "type": "csv",
+      "connection": "hald-small-train.txt",
+      "direction": "import"
+    },
+    "myValidSrc": {
+      "comment": "Binds the dataset contained within hald-small-valid.txt to the myValidSrc datasource. This dataset represents the validation partition.",
+      "type": "csv",
+      "connection": "hald-small-valid.txt",
+      "direction": "import"
+    }
+  },
+  "datasets": {
+    "comment": "Datasets are bound to an entry in the 'datasources' section",
+    "myTrainData": {
+      "comment": "Binds the myTrainSrc datasource to the myTrainData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myTrainSrc",
+      "targetCol": "Y"
+    },
+    "myValidData": {
+      "comment": "Binds the myValidSrc datasource to the myValidData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myValidSrc",
+      "targetCol": "Y"
+    }
+  },
+  "estimator": {
+    "comment": "Defines an estimator, including the type, algorithm and parameters. Only 1 estimator may appear in a RASON model.",
+    "knnpEstimator": {
+      "comment": "Defines the knnpEstimator as a k-nearest neighbors regression estimator. For a complete list of all algorithm options, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "type": "regression",
+      "algorithm": "nearestNeighbors"
+    }
+  },
+  "actions": {
+    "comment": "Define actions, evaluations, and data to be used by the estimator created above.",
+    "knnpModel": {
+      "comment": "Fits a model to the myTrainData dataset using the knnpEstimator estimator. The 'export' parameter exports the fitted model to a JSON file which is reused in the k-Nearest Neighbors Scoring example (Examples - Data Science - Scoring -- Scoring Using Nearest Neighbors Regression - Fitted Model: JSON). For a complete list of all available evaluations, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "trainData": "myTrainData",
+      "estimator": "knnpEstimator",
+      "export": "json",
+      "action": "fit",
+      "evaluations": [ "fittedModelJson" ]
+    },
+    "trainScore": {
+      "comment": "Uses the fitted model knnpModel to score the training dataset, myTrainData (the training partition) using the 'action':'predict' parameter. Various parameters are set and evaluations are requested in the results. For a complete list of all available evaluations, see the Regression - Common Parameters section in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "myTrainData",
+      "fittedModel": "knnpModel",
+      "action": "predict",
+      "parameters": {
+        "numNeighbors": 3,
+        "includeTies": true,
+        "stable": true,
+        "weightingScheme": "INVERSE_DISTANCE"
+      },
+      "evaluations": [ "prediction", "sse", "ss", "sst", "mse", "rmse", "mad", "r2", "aoc" ]
+    },
+    "validScore": {
+      "comment": "Uses the fitted model knnpModel to score the validation dataset, myValidData (the validation partition) using the 'action':'predict' parameter. Various parameters are set and evaluations are requested in the results. Note that the evaluation, metrics, returns SSE, SS, SST, MSE, RMSE, MAD, R2 and AOC. For a complete list of all available evaluations, see the Regression - Common Parameters section chapter in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "myValidData",
+      "fittedModel": "knnpModel",
+      "action": "predict",
+      "parameters": {
+        "numNeighbors": 3,
+        "includeTies": true,
+        "stable": true,
+        "weightingScheme": "INVERSE_DISTANCE"
+      },
+      "evaluations": [ "prediction", "metrics" ]
+    }
+  }
+}

package/examples/datamining/Regression/NeuralNetwork.json

@@ -0,0 +1,83 @@
+{
+  "modelName": "NeuralNetworkRegression",
+  "modelDescription": "Regression Neural Network: This RASON model example generates a fitted model and POSTs the model to the RASON server. Then uses the fitted model to score both the training and validation partitions. This example uses two outside data sources, hald-small-train.txt and hald-small-valid.txt. Both may be downloaded by clicking the RASON Examples Download icon (download cloud icon). To solve this model, you must first POST the model to the RASON Server by clicking the POST icon (upload cloud icon on the left). Afterwards click Choose Files (on the Properties pane on the right), browse to the location of the downloaded examples, and select both hald-small-train.txt and hald-small-valid.txt.  Then click the down arrow next to the Play icon and click Solve. For more information on POSTing and solving example models, click Help - Defining Your Data Science Model.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Define datasources to use with this model, in the datasources section. Typically these bind to external data sources due to data volume.",
+    "myTrainSrc": {
+      "comment": "Binds the dataset contained within hald-small-train.txt to the myTrainSrc datasource. This dataset represents the training partition.",
+      "type": "csv",
+      "connection": "hald-small-train.txt",
+      "direction": "import"
+    },
+    "myValidSrc": {
+      "comment": "Binds the dataset contained within hald-small-valid.txt to the myValidSrc datasource. This dataset represents the validation partition.",
+      "type": "csv",
+      "connection": "hald-small-valid.txt",
+      "direction": "import"
+    }
+  },
+  "datasets": {
+    "comment": "Datasets are bound to an entry in the 'datasources' section",
+    "myTrainData": {
+      "comment": "Binds the myTrainSrc datasource to the myTrainData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myTrainSrc",
+      "targetCol": "Y"
+    },
+    "myValidData": {
+      "comment": "Binds the myValidSrc datasource to the myValidData dataset. Note that the target variable is set to the Y variable using the 'targetCol' property.",
+      "binding": "myValidSrc",
+      "targetCol": "Y"
+    }
+  },
+  "estimator": {
+    "comment": "Defines an estimator, including the type, algorithm and parameters. Only 1 estimator may appear in a RASON model.",
+    "nnpEstimator": {
+      "comment": "Defines the nnpEstimator as a regression neural network estimator; various options are set. For a complete list of all algorithm options, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "type": "regression",
+      "algorithm": "neuralNetwork",
+      "parameters": {
+        "numNeurons": [ 4 ],
+        "numEpochs": 500,
+        "errorTolerance": 0.01,
+        "weightDecay": 0.0,
+        "learningRate": 0.5,
+        "WeightMomentum": 0.0,
+        "hiddenLayerActivation": "LOGISTIC_SIGMOID",
+        "outputLayerActivation": "TANH",
+        "weightInitSeed": 123,
+        "learningOrder": "RANDOM",
+        "learningOrderSeed": 1234,
+        "dataForErrorComputation": "ONLY_TRAIN",
+        "maxNumEpochsWithNoImprovement": 100,
+        "minRelativeErrorChange": 1E-05,
+        "maxTrainingTimeSeconds": 5
+      }
+    }
+  },
+  "actions": {
+    "comment": "Define actions, evaluations, and data to be used by the estimator created above.",
+    "nnpModel": {
+      "comment": "Fits a model to the myTrainData dataset using the nnpEstimator estimator. The 'export' parameter exports the fitted model to a JSON file which may be called to score new data later. Various evaluations are requested in the results including the training log, the neuron weights, etc. For a complete list of all available evaluations, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.)",
+      "trainData": "myTrainData",
+      "estimator": "nnpEstimator",
+      "export": "json",
+      "action": "fit",
+      "evaluations": [ "trainingLog", "neuronWeights", "numEpochsUsed", "trainingTime", "stoppingReason", "partitionCausedStopping" ]
+    },
+    "trainScore": {
+      "comment": "Uses the fitted model nnpModel to score the training dataset, myTrainData using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values, and the sse, ss, sst, mse, rmse, mad and r2 statistics. For a complete list of all available evaluations, see the Regression - Common Parameters section in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "myTrainData",
+      "fittedModel": "nnpModel",
+      "action": "predict",
+      "evaluations": [ "prediction", "sse", "ss", "sst", "mse", "rmse", "mad", "r2", "aoc" ]
+    },
+    "validScore": {
+      "comment": "Uses the fitted model nnpModel to score the validation dataset, myvalidData using the 'action':'predict' parameter. Various evaluations are requested in the results such as the predicted values and metrics (sse, ss, sst, mse, rmse, mad, aoc and r2). For a complete list of all available evaluations, see the Regression - Common Parameters section in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "myValidData",
+      "fittedModel": "nnpModel",
+      "action": "predict",
+      "evaluations": [ "prediction", "metrics" ]
+    }
+  }
+}

package/examples/datamining/Scoring/JSONClassifier.json

@@ -0,0 +1,63 @@
+{
+  "modelName": "JSONClassifier",
+  "modelDescription": "This RASON model demonstrates how to score new data using a previously fit, fitted model. In order to solve this model you must first run the Nearest Neighbor Classification model (Examples -- Data Science -- Classification -- NearestNeighbors.json) which generates and POSTs the fitted model used in this example to the RASON server. In addition, this model also imports two datasources, hald-small-binary.txt and hald-small-binary-txt. Both can be downloaded by clicking the RASON Examples download icon. To attach the datasources to this example, click Choose Files (in the Properties pane on the right), browse to the location of the downloaded files, select both files, then click Open.  Afterwards, click the POST icon to post the model with datafiles attached, then click the down arrow next to the Play icon and select Solve.",
+  "modelType": "datamining",
+  "datasources": {
+    "comment": "Attach both data sources by clicking Download Rason Example Data (on the ribbon) and then Choose Files (on the right).",
+    "labeledDataSrc": {
+      "comment": "This is the labeled data source.",
+      "type": "csv",
+      "connection": "hald-small-binary.txt",
+      "direction": "import"
+    },
+    "unlabeledDataSrc": {
+      "comment": "This is the unlabeled data source.",
+      "type": "csv",
+      "connection": "hald-small-binary-score.txt",
+      "direction": "import"
+    }
+  },
+  "datasets": {
+    "comment": "This example illustrates how to score both labeled and unlabeled data but both are not necessary. Note that the Y variable is selected as the output variable, or the target variable, in both datasets.",
+    "labeledData": {
+      "comment": "Binds the labeled data source, labeledDataSrc, to the labeledData dataset.",
+      "binding": "labeledDataSrc",
+      "targetCol": "Y"
+    },
+    "unlabeledData": {
+      "comment": "Binds the unlabeled data source, unlabeledDataSrc, to the unlabeledData dataset.",
+      "binding": "unlabeledDataSrc"
+    }
+  },
+  "fittedModel": {
+    "comment": "Use this section to bind a fitted model to fittedModel. This step is not optional.",
+    "knnModel": {
+      "comment": "Binds the fitted model, NearestNeighborsClassification, residing on the RASON Server to the fittedModel, knnModel.  Recall that the fitted model was generated and POSTed to the RASON Server when the NearestNeighborsClassification example was solved (Examples - Data Science - Classification -- NeartestNeighbors.json).",
+      "modelName": "NearestNeighborsClassification"
+    }
+  },
+  "actions": {
+    "labeledDataPrediction": {
+      "comment": "Scores data in hald-small-binary.txt using the fitted model, NearestNeighborsClassification, and asks for the results. For a complete list of all available evaluations and parameters, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "labeledData",
+      "fittedModel": "knnModel",
+      "action": "predict",
+      "parameters": {
+        "successClass": "1",
+        "successProbability": 0.6
+      },
+      "evaluations": [ "prediction", "posteriorProbability", "confusingMatrix", "metrics" ]
+    },
+    "unlabeledDataPrediction": {
+      "comment": "Scores data in hald-small-binary-score.txt using the fitted model, NearestNeighborsClassification, and asks for the results. For a complete list of all available evaluations and parameters, see the RASON Data Science Components chapter in the RASON Reference Guide. (Click the Help tab to download.) ",
+      "data": "unlabeledData",
+      "fittedModel": "knnModel",
+      "action": "predict",
+      "parameters": {
+        "successClass": "1",
+        "successProbability": 0.6
+      },
+      "evaluations": [ "prediction", "posteriorProbability" ]
+    }
+  }
+}