{
  "$schema": "https://json-schema.org/draft/2020-12/schema",
  "title": "ML regression and causal PLR task (mlr3 backend)",
  "type": "object",
  "properties": {
    "schema_version": {
      "const": 1
    },
    "task_type": {
      "const": "regression"
    },
    "target": {
      "type": "string",
      "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
    },
    "features": {
      "type": "array",
      "items": {
        "type": "string",
        "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
      },
      "minItems": 1,
      "maxItems": 64,
      "uniqueItems": true
    },
    "id_column": {
      "type": "string",
      "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
    },
    "split": {
      "type": "object",
      "properties": {
        "column": {
          "type": "string",
          "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
        },
        "train": {
          "type": "string",
          "pattern": "^[A-Za-z][A-Za-z0-9_-]{0,31}$"
        },
        "val": {
          "type": "string",
          "pattern": "^[A-Za-z][A-Za-z0-9_-]{0,31}$"
        },
        "test": {
          "type": "string",
          "pattern": "^[A-Za-z][A-Za-z0-9_-]{0,31}$"
        }
      },
      "required": [
        "column",
        "train",
        "val",
        "test"
      ],
      "additionalProperties": false
    },
    "categorical_features": {
      "type": "array",
      "items": {
        "type": "string",
        "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
      },
      "minItems": 0,
      "maxItems": 64,
      "uniqueItems": true
    },
    "missing": {
      "type": "object",
      "properties": {
        "policy": {
          "enum": [
            "drop",
            "train_median_mode"
          ]
        },
        "sentinels": {
          "type": "array",
          "items": {
            "type": "number"
          },
          "minItems": 0,
          "maxItems": 8,
          "uniqueItems": true
        }
      },
      "required": [
        "policy",
        "sentinels"
      ],
      "additionalProperties": false
    },
    "models": {
      "type": "array",
      "items": {
        "oneOf": [
          {
            "type": "object",
            "properties": {
              "id": {
                "type": "string",
                "pattern": "^[a-z][a-z0-9_]{0,31}$"
              },
              "learner": {
                "const": "regr.rpart"
              },
              "parameters": {
                "type": "object",
                "properties": {
                  "cp": {
                    "type": "number",
                    "minimum": 0,
                    "maximum": 1
                  },
                  "maxdepth": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 30
                  },
                  "minsplit": {
                    "type": "integer",
                    "minimum": 2,
                    "maximum": 10000
                  },
                  "minbucket": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 5000
                  }
                },
                "required": [],
                "additionalProperties": false
              }
            },
            "required": [
              "id",
              "learner",
              "parameters"
            ],
            "additionalProperties": false
          },
          {
            "type": "object",
            "properties": {
              "id": {
                "type": "string",
                "pattern": "^[a-z][a-z0-9_]{0,31}$"
              },
              "learner": {
                "const": "regr.ranger"
              },
              "parameters": {
                "type": "object",
                "properties": {
                  "num.trees": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 1000
                  },
                  "mtry": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 64
                  },
                  "min.node.size": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 10000
                  },
                  "max.depth": {
                    "type": "integer",
                    "minimum": 1,
                    "maximum": 30
                  },
                  "sample.fraction": {
                    "type": "number",
                    "minimum": 0.1,
                    "maximum": 1
                  },
                  "replace": {
                    "type": "boolean"
                  }
                },
                "required": [],
                "additionalProperties": false
              }
            },
            "required": [
              "id",
              "learner",
              "parameters"
            ],
            "additionalProperties": false
          }
        ]
      },
      "minItems": 1,
      "maxItems": 4
    },
    "selection_metric": {
      "enum": [
        "rmse",
        "mae"
      ]
    },
    "seed": {
      "type": "integer",
      "minimum": 1,
      "maximum": 2147483000
    },
    "row_limits": {
      "type": "object",
      "properties": {
        "train": {
          "type": "integer",
          "minimum": 10,
          "maximum": 20000
        },
        "val": {
          "type": "integer",
          "minimum": 5,
          "maximum": 5000
        },
        "test": {
          "type": "integer",
          "minimum": 5,
          "maximum": 5000
        }
      },
      "required": [
        "train",
        "val",
        "test"
      ],
      "additionalProperties": false
    },
    "explanation": {
      "type": "object",
      "properties": {
        "features": {
          "type": "array",
          "items": {
            "type": "string",
            "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
          },
          "minItems": 1,
          "maxItems": 5,
          "uniqueItems": true
        },
        "background_rows": {
          "type": "integer",
          "minimum": 5,
          "maximum": 200
        },
        "validation_rows": {
          "type": "integer",
          "minimum": 5,
          "maximum": 200
        },
        "permutation_repeats": {
          "type": "integer",
          "minimum": 1,
          "maximum": 10
        },
        "pdp_points": {
          "type": "integer",
          "minimum": 2,
          "maximum": 20
        },
        "methods": {
          "type": "array",
          "items": {
            "enum": [
              "permutation",
              "pdp",
              "ice",
              "shap"
            ]
          },
          "minItems": 1,
          "maxItems": 4,
          "uniqueItems": true
        },
        "shap_permutations": {
          "type": "integer",
          "minimum": 2,
          "maximum": 32
        },
        "shap_rows": {
          "type": "integer",
          "minimum": 1,
          "maximum": 20
        }
      },
      "required": [
        "features",
        "background_rows",
        "validation_rows",
        "permutation_repeats",
        "pdp_points"
      ],
      "additionalProperties": false
    },
    "requirements": {
      "type": "string",
      "maxLength": 1000
    },
    "target_unit": {
      "type": "string",
      "maxLength": 32
    },
    "unseen_categories": {
      "enum": [
        "error",
        "train_mode"
      ],
      "default": "error"
    },
    "analysis": {
      "enum": [
        "prediction",
        "causal_plr"
      ],
      "default": "prediction"
    },
    "target_definition": {
      "type": "object",
      "properties": {
        "sample_unit": {
          "type": "string",
          "maxLength": 120
        },
        "averaging_period": {
          "type": "string",
          "maxLength": 120
        },
        "prediction_time": {
          "type": "string",
          "maxLength": 120
        }
      },
      "required": [],
      "additionalProperties": false
    },
    "excluded_features": {
      "type": "array",
      "items": {
        "type": "string",
        "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
      },
      "minItems": 0,
      "maxItems": 64,
      "uniqueItems": true
    },
    "feature_selection": {
      "type": "object",
      "properties": {
        "mode": {
          "enum": [
            "manual",
            "train_filter"
          ]
        },
        "max_missing_fraction": {
          "type": "number",
          "minimum": 0,
          "maximum": 1
        },
        "min_unique": {
          "type": "integer",
          "minimum": 2,
          "maximum": 100
        },
        "max_abs_correlation": {
          "type": "number",
          "minimum": 0,
          "maximum": 1
        }
      },
      "required": [
        "mode"
      ],
      "additionalProperties": false
    },
    "causal": {
      "type": "object",
      "properties": {
        "treatment": {
          "type": "string",
          "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
        },
        "treatment_unit": {
          "type": "string",
          "maxLength": 60
        },
        "treatment_scale": {
          "type": "number",
          "exclusiveMinimum": 0,
          "maximum": 1000000000000.0
        },
        "estimand": {
          "const": "total_effect_plr_slope"
        },
        "adjustment_set": {
          "type": "array",
          "items": {
            "type": "string",
            "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
          },
          "minItems": 1,
          "maxItems": 28,
          "uniqueItems": true
        },
        "folds": {
          "type": "integer",
          "minimum": 2,
          "maximum": 10
        },
        "resampling": {
          "type": "object",
          "properties": {
            "mode": {
              "enum": [
                "iid",
                "group"
              ]
            },
            "group_column": {
              "type": "string",
              "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
            }
          },
          "required": [
            "mode"
          ],
          "additionalProperties": false
        },
        "assumptions": {
          "type": "object",
          "properties": {
            "causal_ordering": {
              "const": true
            },
            "conditional_exchangeability": {
              "const": true
            },
            "consistency": {
              "const": true
            },
            "overlap": {
              "const": true
            },
            "partially_linear_effect": {
              "const": true
            },
            "independent_units_or_clusters": {
              "const": true
            }
          },
          "required": [
            "causal_ordering",
            "conditional_exchangeability",
            "consistency",
            "overlap",
            "partially_linear_effect",
            "independent_units_or_clusters"
          ],
          "additionalProperties": false
        },
        "dag": {
          "type": "object",
          "properties": {
            "nodes": {
              "type": "array",
              "items": {
                "type": "string",
                "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
              },
              "minItems": 2,
              "maxItems": 30,
              "uniqueItems": true
            },
            "edges": {
              "type": "array",
              "items": {
                "type": "array",
                "items": {
                  "type": "string",
                  "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
                },
                "minItems": 2,
                "maxItems": 2
              },
              "minItems": 0,
              "maxItems": 120,
              "uniqueItems": true
            },
            "latent": {
              "type": "array",
              "items": {
                "type": "string",
                "pattern": "^[A-Za-z_][A-Za-z0-9_.-]{0,95}$"
              },
              "minItems": 0,
              "maxItems": 20,
              "uniqueItems": true
            }
          },
          "required": [
            "nodes",
            "edges"
          ],
          "additionalProperties": false
        }
      },
      "required": [
        "treatment",
        "treatment_unit",
        "treatment_scale",
        "estimand",
        "adjustment_set",
        "folds",
        "resampling",
        "assumptions",
        "dag"
      ],
      "additionalProperties": false
    },
    "output_files": {
      "type": "object",
      "properties": {
        "target_contract": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "prepared_data": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "preprocessing": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "feature_selection": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "candidates": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "model": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "predictions": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "metrics": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "permutation": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "ice": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "pdp": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "importance_plot": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "dependence_plot": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "shap": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "dag": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "dag_plot": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "causal_effect": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "causal_residuals": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "causal_folds": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "causal_models": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "provenance": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "reproducibility": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "status": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        },
        "artifacts": {
          "type": "string",
          "minLength": 1,
          "maxLength": 240
        }
      },
      "required": [],
      "additionalProperties": false
    },
    "seeds": {
      "type": "object",
      "properties": {
        "sampling": {
          "type": "integer",
          "minimum": 1,
          "maximum": 2147483000
        },
        "model": {
          "type": "integer",
          "minimum": 1,
          "maximum": 2147483000
        },
        "explanation": {
          "type": "integer",
          "minimum": 1,
          "maximum": 2147483000
        },
        "causal": {
          "type": "integer",
          "minimum": 1,
          "maximum": 2147483000
        }
      },
      "additionalProperties": false
    }
  },
  "required": [
    "schema_version",
    "task_type",
    "target",
    "features",
    "id_column",
    "split",
    "categorical_features",
    "missing",
    "models",
    "seed",
    "row_limits"
  ],
  "additionalProperties": false,
  "allOf": [
    {
      "if": {
        "properties": {
          "analysis": {
            "const": "causal_plr"
          }
        },
        "required": [
          "analysis"
        ]
      },
      "then": {
        "required": [
          "causal"
        ]
      },
      "else": {
        "required": [
          "selection_metric",
          "explanation"
        ]
      }
    }
  ]
}