@mat3ra/esse 2024.3.25-2 → 2024.3.25-3

package/dist/js/schema/software_directory/ml/unit/processing/data_transformation.json

@@ -0,0 +1,334 @@
+{
+    "$id": "software-directory/ml/unit/processing/data-transformation",
+    "oneOf": [
+        {
+            "$schema": "http://json-schema.org/draft-07/schema#",
+            "title": "scale and reduce unit schema",
+            "type": "object",
+            "required": [
+                "flowchartId",
+                "inputData",
+                "operation",
+                "operationType",
+                "type"
+            ],
+            "additionalProperties": true,
+            "properties": {
+                "operation": {
+                    "type": "string",
+                    "enum": [
+                        "data_transformation"
+                    ],
+                    "description": "Contains information about the operation used."
+                },
+                "operationType": {
+                    "type": "string",
+                    "enum": [
+                        "scale_and_reduce"
+                    ],
+                    "description": "Contains information about the specific type of the operation used."
+                },
+                "inputData": {
+                    "type": "object",
+                    "required": [
+                        "scaler"
+                    ],
+                    "description": "unit input (type to be specified by the child units)",
+                    "properties": {
+                        "scaler": {
+                            "description": "type of scaler to be applied",
+                            "type": "string",
+                            "default": "standard_scaler",
+                            "enum": [
+                                "standard_scaler"
+                            ]
+                        },
+                        "perFeature": {
+                            "description": "per-feature scaling data",
+                            "type": "array",
+                            "items": {
+                                "type": "object",
+                                "required": [
+                                    "name",
+                                    "scale"
+                                ],
+                                "properties": {
+                                    "variance": {
+                                        "description": "variance in original training data",
+                                        "type": "number"
+                                    },
+                                    "mean": {
+                                        "description": "mean value of the original training data",
+                                        "type": "number"
+                                    },
+                                    "scale": {
+                                        "description": "scale multiplier for this feature/property",
+                                        "type": "number"
+                                    },
+                                    "name": {
+                                        "description": "feature/property name in 'flattened' format",
+                                        "type": "string"
+                                    }
+                                }
+                            }
+                        }
+                    }
+                },
+                "type": {
+                    "enum": [
+                        "processing"
+                    ],
+                    "description": "type of the unit",
+                    "type": "string"
+                },
+                "_id": {
+                    "type": "string",
+                    "description": "entity identity"
+                },
+                "isDraft": {
+                    "type": "boolean"
+                },
+                "name": {
+                    "description": "name of the unit. e.g. pw_scf",
+                    "type": "string"
+                },
+                "status": {
+                    "description": "Status of the unit.",
+                    "type": "string",
+                    "enum": [
+                        "idle",
+                        "active",
+                        "warning",
+                        "error",
+                        "finished"
+                    ]
+                },
+                "head": {
+                    "description": "Whether this unit is the first one to be executed.",
+                    "type": "boolean"
+                },
+                "flowchartId": {
+                    "description": "Identity of the unit in the workflow. Used to trace the execution flow of the workflow.",
+                    "type": "string"
+                },
+                "next": {
+                    "description": "Next unit's flowchartId. If empty, the current unit is the last.",
+                    "type": "string"
+                },
+                "enableRender": {
+                    "description": "Whether Rupy should attempt to use Jinja templating to add context variables into the unit",
+                    "type": "boolean"
+                },
+                "context": {
+                    "type": "object"
+                },
+                "slug": {
+                    "description": "entity slug",
+                    "type": "string"
+                },
+                "systemName": {
+                    "type": "string"
+                },
+                "consistencyChecks": {
+                    "type": "array",
+                    "items": {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "consistency check",
+                        "type": "object",
+                        "description": "The output of consistency checks performed on data adhering to JSON schema, but inconsistent with scientific or logical rules, to show problems in UI.",
+                        "required": [
+                            "key",
+                            "name",
+                            "severity",
+                            "message"
+                        ],
+                        "properties": {
+                            "key": {
+                                "type": "string",
+                                "description": "Key of the property of the entity on which the consistency check is performed in Mongo dot notation, e.g. 'basis.coordinates.1'"
+                            },
+                            "name": {
+                                "enum": [
+                                    "default",
+                                    "atomsTooClose",
+                                    "atomsOverlap"
+                                ],
+                                "description": "Name of the consistency check that is performed, which is listed in an enum."
+                            },
+                            "severity": {
+                                "enum": [
+                                    "info",
+                                    "warning",
+                                    "error"
+                                ],
+                                "description": "Severity level of the problem, which is used in UI to differentiate."
+                            },
+                            "message": {
+                                "type": "string",
+                                "description": "Message generated by the consistency check describing the problem."
+                            }
+                        }
+                    }
+                },
+                "schemaVersion": {
+                    "description": "entity's schema version. Used to distinct between different schemas.",
+                    "type": "string",
+                    "default": "2022.8.16"
+                },
+                "isDefault": {
+                    "description": "Identifies that entity is defaultable",
+                    "type": "boolean",
+                    "default": false
+                },
+                "preProcessors": {
+                    "description": "names of the pre-processors for this calculation",
+                    "type": "array",
+                    "items": {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "runtime item schema",
+                        "oneOf": [
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "name result schema",
+                                "type": "object",
+                                "properties": {
+                                    "name": {
+                                        "description": "The name of this item. e.g. scf_accuracy",
+                                        "type": "string"
+                                    }
+                                },
+                                "required": [
+                                    "name"
+                                ]
+                            },
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "RuntimeItemString",
+                                "description": "name of runtime item in shortened notation",
+                                "type": "string"
+                            }
+                        ]
+                    }
+                },
+                "postProcessors": {
+                    "description": "names of the post-processors for this calculation",
+                    "type": "array",
+                    "items": {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "runtime item schema",
+                        "oneOf": [
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "name result schema",
+                                "type": "object",
+                                "properties": {
+                                    "name": {
+                                        "description": "The name of this item. e.g. scf_accuracy",
+                                        "type": "string"
+                                    }
+                                },
+                                "required": [
+                                    "name"
+                                ]
+                            },
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "RuntimeItemString",
+                                "description": "name of runtime item in shortened notation",
+                                "type": "string"
+                            }
+                        ]
+                    }
+                },
+                "monitors": {
+                    "description": "names of the monitors for this calculation",
+                    "type": "array",
+                    "items": {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "runtime item schema",
+                        "oneOf": [
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "name result schema",
+                                "type": "object",
+                                "properties": {
+                                    "name": {
+                                        "description": "The name of this item. e.g. scf_accuracy",
+                                        "type": "string"
+                                    }
+                                },
+                                "required": [
+                                    "name"
+                                ]
+                            },
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "RuntimeItemString",
+                                "description": "name of runtime item in shortened notation",
+                                "type": "string"
+                            }
+                        ]
+                    }
+                },
+                "results": {
+                    "description": "names of the results for this calculation",
+                    "type": "array",
+                    "items": {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "runtime item schema",
+                        "oneOf": [
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "name result schema",
+                                "type": "object",
+                                "properties": {
+                                    "name": {
+                                        "description": "The name of this item. e.g. scf_accuracy",
+                                        "type": "string"
+                                    }
+                                },
+                                "required": [
+                                    "name"
+                                ]
+                            },
+                            {
+                                "$schema": "http://json-schema.org/draft-07/schema#",
+                                "title": "RuntimeItemString",
+                                "description": "name of runtime item in shortened notation",
+                                "type": "string"
+                            }
+                        ]
+                    }
+                },
+                "tags": {
+                    "description": "entity tags",
+                    "type": "array",
+                    "items": {
+                        "type": "string"
+                    }
+                },
+                "statusTrack": {
+                    "type": "array",
+                    "items": {
+                        "type": "object",
+                        "required": [
+                            "trackedAt",
+                            "status"
+                        ],
+                        "properties": {
+                            "trackedAt": {
+                                "type": "number"
+                            },
+                            "status": {
+                                "type": "string"
+                            },
+                            "repetition": {
+                                "type": "number"
+                            }
+                        }
+                    }
+                }
+            }
+        }
+    ]
+}

package/dist/js/schema/software_directory/ml/unit/processing/feature_selection/filter_based.json

@@ -0,0 +1,305 @@
+{
+    "$id": "software-directory/ml/unit/processing/feature-selection/filter-based",
+    "$schema": "http://json-schema.org/draft-07/schema#",
+    "title": "filter-based feature selection unit schema",
+    "type": "object",
+    "required": [
+        "flowchartId",
+        "inputData",
+        "operation",
+        "operationType",
+        "type"
+    ],
+    "additionalProperties": true,
+    "properties": {
+        "operation": {
+            "type": "string",
+            "enum": [
+                "feature_selection"
+            ],
+            "description": "Contains information about the operation used."
+        },
+        "operationType": {
+            "type": "string",
+            "enum": [
+                "filter_based"
+            ],
+            "description": "Contains information about the specific type of the operation used."
+        },
+        "inputData": {
+            "type": "object",
+            "required": [
+                "nFeatures",
+                "algorithm"
+            ],
+            "description": "unit input (type to be specified by the child units)",
+            "properties": {
+                "nFeatures": {
+                    "description": "number of features to select for model training. If equal to 0, will use all available features",
+                    "type": "number"
+                },
+                "algorithm": {
+                    "description": "feature selection algorithm following sklearn.feature_selection",
+                    "type": "string",
+                    "enum": [
+                        "f_regression"
+                    ]
+                }
+            }
+        },
+        "type": {
+            "enum": [
+                "processing"
+            ],
+            "description": "type of the unit",
+            "type": "string"
+        },
+        "_id": {
+            "type": "string",
+            "description": "entity identity"
+        },
+        "isDraft": {
+            "type": "boolean"
+        },
+        "name": {
+            "description": "name of the unit. e.g. pw_scf",
+            "type": "string"
+        },
+        "status": {
+            "description": "Status of the unit.",
+            "type": "string",
+            "enum": [
+                "idle",
+                "active",
+                "warning",
+                "error",
+                "finished"
+            ]
+        },
+        "head": {
+            "description": "Whether this unit is the first one to be executed.",
+            "type": "boolean"
+        },
+        "flowchartId": {
+            "description": "Identity of the unit in the workflow. Used to trace the execution flow of the workflow.",
+            "type": "string"
+        },
+        "next": {
+            "description": "Next unit's flowchartId. If empty, the current unit is the last.",
+            "type": "string"
+        },
+        "enableRender": {
+            "description": "Whether Rupy should attempt to use Jinja templating to add context variables into the unit",
+            "type": "boolean"
+        },
+        "context": {
+            "type": "object"
+        },
+        "slug": {
+            "description": "entity slug",
+            "type": "string"
+        },
+        "systemName": {
+            "type": "string"
+        },
+        "consistencyChecks": {
+            "type": "array",
+            "items": {
+                "$schema": "http://json-schema.org/draft-07/schema#",
+                "title": "consistency check",
+                "type": "object",
+                "description": "The output of consistency checks performed on data adhering to JSON schema, but inconsistent with scientific or logical rules, to show problems in UI.",
+                "required": [
+                    "key",
+                    "name",
+                    "severity",
+                    "message"
+                ],
+                "properties": {
+                    "key": {
+                        "type": "string",
+                        "description": "Key of the property of the entity on which the consistency check is performed in Mongo dot notation, e.g. 'basis.coordinates.1'"
+                    },
+                    "name": {
+                        "enum": [
+                            "default",
+                            "atomsTooClose",
+                            "atomsOverlap"
+                        ],
+                        "description": "Name of the consistency check that is performed, which is listed in an enum."
+                    },
+                    "severity": {
+                        "enum": [
+                            "info",
+                            "warning",
+                            "error"
+                        ],
+                        "description": "Severity level of the problem, which is used in UI to differentiate."
+                    },
+                    "message": {
+                        "type": "string",
+                        "description": "Message generated by the consistency check describing the problem."
+                    }
+                }
+            }
+        },
+        "schemaVersion": {
+            "description": "entity's schema version. Used to distinct between different schemas.",
+            "type": "string",
+            "default": "2022.8.16"
+        },
+        "isDefault": {
+            "description": "Identifies that entity is defaultable",
+            "type": "boolean",
+            "default": false
+        },
+        "preProcessors": {
+            "description": "names of the pre-processors for this calculation",
+            "type": "array",
+            "items": {
+                "$schema": "http://json-schema.org/draft-07/schema#",
+                "title": "runtime item schema",
+                "oneOf": [
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "name result schema",
+                        "type": "object",
+                        "properties": {
+                            "name": {
+                                "description": "The name of this item. e.g. scf_accuracy",
+                                "type": "string"
+                            }
+                        },
+                        "required": [
+                            "name"
+                        ]
+                    },
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "RuntimeItemString",
+                        "description": "name of runtime item in shortened notation",
+                        "type": "string"
+                    }
+                ]
+            }
+        },
+        "postProcessors": {
+            "description": "names of the post-processors for this calculation",
+            "type": "array",
+            "items": {
+                "$schema": "http://json-schema.org/draft-07/schema#",
+                "title": "runtime item schema",
+                "oneOf": [
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "name result schema",
+                        "type": "object",
+                        "properties": {
+                            "name": {
+                                "description": "The name of this item. e.g. scf_accuracy",
+                                "type": "string"
+                            }
+                        },
+                        "required": [
+                            "name"
+                        ]
+                    },
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "RuntimeItemString",
+                        "description": "name of runtime item in shortened notation",
+                        "type": "string"
+                    }
+                ]
+            }
+        },
+        "monitors": {
+            "description": "names of the monitors for this calculation",
+            "type": "array",
+            "items": {
+                "$schema": "http://json-schema.org/draft-07/schema#",
+                "title": "runtime item schema",
+                "oneOf": [
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "name result schema",
+                        "type": "object",
+                        "properties": {
+                            "name": {
+                                "description": "The name of this item. e.g. scf_accuracy",
+                                "type": "string"
+                            }
+                        },
+                        "required": [
+                            "name"
+                        ]
+                    },
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "RuntimeItemString",
+                        "description": "name of runtime item in shortened notation",
+                        "type": "string"
+                    }
+                ]
+            }
+        },
+        "results": {
+            "description": "names of the results for this calculation",
+            "type": "array",
+            "items": {
+                "$schema": "http://json-schema.org/draft-07/schema#",
+                "title": "runtime item schema",
+                "oneOf": [
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "name result schema",
+                        "type": "object",
+                        "properties": {
+                            "name": {
+                                "description": "The name of this item. e.g. scf_accuracy",
+                                "type": "string"
+                            }
+                        },
+                        "required": [
+                            "name"
+                        ]
+                    },
+                    {
+                        "$schema": "http://json-schema.org/draft-07/schema#",
+                        "title": "RuntimeItemString",
+                        "description": "name of runtime item in shortened notation",
+                        "type": "string"
+                    }
+                ]
+            }
+        },
+        "tags": {
+            "description": "entity tags",
+            "type": "array",
+            "items": {
+                "type": "string"
+            }
+        },
+        "statusTrack": {
+            "type": "array",
+            "items": {
+                "type": "object",
+                "required": [
+                    "trackedAt",
+                    "status"
+                ],
+                "properties": {
+                    "trackedAt": {
+                        "type": "number"
+                    },
+                    "status": {
+                        "type": "string"
+                    },
+                    "repetition": {
+                        "type": "number"
+                    }
+                }
+            }
+        }
+    }
+}