{
  "format": "dataset-manifest/1",
  "slug": "database-script-for-high-performance-neural-network-inference-framework",
  "version": 3,
  "seed": "database-script-for-high-performance-neural-network-inference-framework:1",
  "spec": {
    "format": "dataset/1",
    "slug": "database-script-for-high-performance-neural-network-inference-framework",
    "name": "Neural Network Performance Insights",
    "tagline": "Inference requests, model performance metrics, and GPU utilization data from a telecom neural network framework.",
    "description": "Neural networks process thousands of inference requests daily, optimizing telecom services and enhancing user experiences. Engineers monitor and adjust these models to ensure they operate efficiently, diving into metrics that reflect how well the systems perform under various loads. At the heart of this operation, user interactions are captured, providing insights into how individuals engage with the neural network, while detailed records of GPU utilization reveal the computational demands of processing these requests. \n\nThe dataset includes several tables: 'inference_requests' logs every incoming request and its attributes, linking to 'models' that detail the neural network architectures in use. Each model is defined by specific 'hyperparameters', which are essential settings influencing the model's performance and efficiency. Performance is ultimately measured through 'performance_metrics', outlining the accuracy and speed of predictions for each model. Additionally, the 'gpu_utilization' table records the resource consumption associated with each inference request, providing context for performance outcomes. Lastly, 'user_interactions' tracks how users engage with the system, adding a layer of behavioral data to the analysis. \n\nThis interconnected structure allows developers to explore questions such as: What configurations yield the best performance? How does GPU utilization impact model efficiency? How do user interactions correlate with model inference times? Analyzing this data aids in optimizing neural network operations, making it an excellent resource for those looking to practice their data handling and analytical skills. Keep in mind that the data is synthetic, designed to mimic real-world scenarios in telecom without revealing any sensitive information.",
    "domain": "telecom",
    "shape": "slowly-changing",
    "useCases": [
      "etl-pipelines",
      "teaching"
    ],
    "keywords": [
      "neural network",
      "telecom",
      "performance metrics",
      "GPU utilization",
      "hyperparameters",
      "inference requests",
      "data analysis",
      "synthetic dataset"
    ],
    "locale": "en",
    "license": "CC0-1.0",
    "version": 1,
    "tables": {
      "inference_requests": {
        "description": "Each row represents an inference request processed by a neural network model, linking to model configurations and performance metrics.",
        "rows": 500,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each inference request."
          },
          "model_id": {
            "kind": "fk",
            "ref": "models.id",
            "description": "Reference to the model used for inference."
          },
          "request_time": {
            "kind": "datetime",
            "from": "2023-01-01",
            "to": "2023-12-31",
            "description": "Timestamp of when the inference request was made.",
            "dayProfile": "waking"
          },
          "input_data_size": {
            "kind": "int",
            "min": 1,
            "max": 10000,
            "description": "Size of input data for the inference in kilobytes."
          },
          "response_time": {
            "kind": "decimal",
            "min": 0,
            "max": 5,
            "scale": 2,
            "description": "Time taken to process the inference request in seconds.",
            "distribution": "normal"
          },
          "success": {
            "kind": "bool",
            "trueRate": 0.95,
            "description": "Indicates if the inference request was successful."
          }
        }
      },
      "models": {
        "description": "This table holds the various neural network models used for processing inference requests, linking to their hyperparameter settings.",
        "rows": 30,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each model."
          },
          "model_name": {
            "kind": "fullName",
            "description": "Name of the neural network model."
          },
          "hyperparameter_id": {
            "kind": "fk",
            "ref": "hyperparameters.id",
            "description": "Reference to the hyperparameter settings for the model."
          },
          "creation_date": {
            "kind": "date",
            "from": "2023-01-01",
            "to": "2023-12-31",
            "description": "Date when the model was created.",
            "trend": "growing"
          },
          "version": {
            "kind": "int",
            "min": 1,
            "max": 10,
            "description": "Version of the model."
          }
        }
      },
      "hyperparameters": {
        "description": "This table contains the hyperparameter settings for the neural network models, which can influence performance.",
        "rows": 10,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each hyperparameter set."
          },
          "learning_rate": {
            "kind": "decimal",
            "min": 0.0001,
            "max": 0.1,
            "scale": 4,
            "description": "Learning rate used for model training.",
            "nullable": true,
            "when": {
              "column": "optimizer",
              "in": [
                "SGD",
                "Adam",
                "RMSprop"
              ]
            }
          },
          "batch_size": {
            "kind": "int",
            "min": 1,
            "max": 128,
            "description": "Batch size used during training.",
            "nullable": true,
            "when": {
              "column": "optimizer",
              "in": [
                "SGD",
                "Adam",
                "RMSprop"
              ]
            }
          },
          "dropout_rate": {
            "kind": "decimal",
            "min": 0,
            "max": 0.9,
            "scale": 2,
            "description": "Dropout rate applied in the model.",
            "nullable": true,
            "when": {
              "column": "optimizer",
              "in": [
                "SGD",
                "Adam",
                "RMSprop"
              ]
            },
            "distribution": "normal"
          },
          "optimizer": {
            "kind": "enum",
            "values": [
              "SGD",
              "Adam",
              "RMSprop"
            ],
            "description": "Optimizer used for training the model.",
            "weights": [
              7,
              2,
              1
            ]
          }
        }
      },
      "gpu_utilization": {
        "description": "This table tracks GPU utilization metrics during inference requests, linking to specific requests.",
        "rows": 600,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each GPU utilization record."
          },
          "inference_request_id": {
            "kind": "fk",
            "ref": "inference_requests.id",
            "description": "Reference to the associated inference request."
          },
          "gpu_id": {
            "kind": "uuid",
            "description": "Identifier for the GPU used."
          },
          "utilization_percentage": {
            "kind": "int",
            "min": 0,
            "max": 100,
            "description": "Percentage of GPU utilization during the inference.",
            "distribution": "high"
          },
          "timestamp": {
            "kind": "datetime",
            "from": "2023-01-01",
            "to": "2023-12-31",
            "description": "Timestamp of the utilization record."
          }
        }
      },
      "performance_metrics": {
        "description": "This table stores performance metrics for each model, allowing for comparative analysis.",
        "rows": 100,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each performance metric record."
          },
          "model_id": {
            "kind": "fk",
            "ref": "models.id",
            "description": "Reference to the model for which metrics are recorded."
          },
          "accuracy": {
            "kind": "decimal",
            "min": 0,
            "max": 1,
            "scale": 4,
            "description": "Accuracy of the model on the validation set.",
            "distribution": "high"
          },
          "latency": {
            "kind": "decimal",
            "min": 0,
            "max": 10,
            "scale": 2,
            "description": "Average latency of the model in seconds.",
            "distribution": "normal"
          },
          "f1_score": {
            "kind": "decimal",
            "min": 0,
            "max": 1,
            "scale": 4,
            "description": "F1 score of the model.",
            "distribution": "high"
          }
        }
      },
      "user_interactions": {
        "description": "This table logs user interactions with the neural network framework, providing context for inference requests.",
        "rows": 200,
        "columns": {
          "id": {
            "kind": "pk",
            "description": "Unique identifier for each user interaction."
          },
          "inference_request_id": {
            "kind": "fk",
            "ref": "inference_requests.id",
            "description": "Link to the associated inference request."
          },
          "user_id": {
            "kind": "uuid",
            "description": "Identifier for the user making the request."
          },
          "interaction_time": {
            "kind": "datetime",
            "from": "2023-01-01",
            "to": "2023-12-31",
            "description": "Timestamp of the user interaction.",
            "dayProfile": "waking",
            "trend": "seasonal"
          },
          "feedback": {
            "kind": "sentence",
            "min": 5,
            "max": 140,
            "description": "User feedback on the inference result."
          }
        }
      }
    },
    "structure": "multi-table",
    "traits": [
      "timestamped",
      "categorical",
      "nullable"
    ],
    "analysisQuestions": [
      "What hyperparameter settings yield the highest model accuracy?",
      "How does GPU utilization vary with different models during inference?",
      "Which models perform best under specific user interaction scenarios?",
      "What is the average response time for inference requests?",
      "Are there any trends in user feedback related to model performance?"
    ],
    "suggestedQueries": [
      {
        "title": "Join Inference Requests and Models",
        "sql": "SELECT ir.id, ir.request_time, m.model_name FROM inference_requests ir JOIN models m ON ir.model_id = m.id;",
        "dialect": "postgres"
      },
      {
        "title": "Average Response Time per Model",
        "sql": "SELECT m.model_name, AVG(ir.response_time) as avg_response_time FROM inference_requests ir JOIN models m ON ir.model_id = m.id GROUP BY m.model_name;",
        "dialect": "postgres"
      },
      {
        "title": "Performance Metrics Overview",
        "sql": "SELECT m.model_name, pm.accuracy, pm.latency FROM performance_metrics pm JOIN models m ON pm.model_id = m.id;",
        "dialect": "postgres"
      },
      {
        "title": "User Feedback Analysis",
        "sql": "SELECT ir.id, ui.feedback FROM inference_requests ir JOIN user_interactions ui ON ir.id = ui.inference_request_id WHERE ui.feedback IS NOT NULL;",
        "dialect": "postgres"
      },
      {
        "title": "GPU Utilization Trends Over Time",
        "sql": "SELECT gpu_id, AVG(utilization_percentage) as avg_utilization, DATE(timestamp) as day FROM gpu_utilization GROUP BY gpu_id, day ORDER BY day;",
        "dialect": "postgres"
      }
    ],
    "usageScenarios": [
      {
        "title": "Data Analysts in Telecom",
        "body": "Data analysts in the telecom industry can leverage this dataset to evaluate the performance of different neural network models. By analyzing inference requests and their associated metrics, they can identify trends and make recommendations for optimizing model performance and improving user experiences."
      },
      {
        "title": "Educators Teaching Data Science",
        "body": "Educators can use this dataset as a teaching tool in data science courses, providing students with real-world scenarios to practice SQL querying and data analysis. The dataset's structure allows students to explore relationships between models, hyperparameters, and user interactions, enhancing their learning experience."
      },
      {
        "title": "Machine Learning Engineers",
        "body": "Machine learning engineers can utilize this dataset to assess the impact of various hyperparameter settings on model performance. By experimenting with different configurations, they can gain insights into how to fine-tune models for optimal results in production environments."
      },
      {
        "title": "Performance Monitoring Teams",
        "body": "Performance monitoring teams can analyze GPU utilization data to ensure resources are being used efficiently during inference requests. By identifying bottlenecks and underutilized resources, they can make informed decisions about infrastructure improvements."
      }
    ]
  },
  "counts": {
    "hyperparameters": 10,
    "models": 30,
    "inference_requests": 500,
    "gpu_utilization": 600,
    "performance_metrics": 100,
    "user_interactions": 200
  },
  "stats": {
    "inference_requests": {
      "rows": 500,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 500,
          "min": 1,
          "max": 500
        },
        "model_id": {
          "nulls": 0,
          "distinct": 30,
          "min": 1,
          "max": 30
        },
        "request_time": {
          "nulls": 0,
          "distinct": 500,
          "min": "2023-01-02 10:10:07",
          "max": "2023-12-31 18:56:23"
        },
        "input_data_size": {
          "nulls": 0,
          "distinct": 489,
          "min": 6,
          "max": 9997
        },
        "response_time": {
          "nulls": 0,
          "distinct": 227,
          "min": 0.54,
          "max": 4.49
        },
        "success": {
          "nulls": 0,
          "distinct": 2,
          "min": "false",
          "max": "true"
        }
      }
    },
    "models": {
      "rows": 30,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 30,
          "min": 1,
          "max": 30
        },
        "model_name": {
          "nulls": 0,
          "distinct": 30,
          "min": "Andre Carter",
          "max": "Yusuf Nakamura"
        },
        "hyperparameter_id": {
          "nulls": 0,
          "distinct": 9,
          "min": 1,
          "max": 10
        },
        "creation_date": {
          "nulls": 0,
          "distinct": 28,
          "min": "2023-02-04",
          "max": "2023-12-30"
        },
        "version": {
          "nulls": 0,
          "distinct": 10,
          "min": 1,
          "max": 10
        }
      }
    },
    "hyperparameters": {
      "rows": 10,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 10,
          "min": 1,
          "max": 10
        },
        "learning_rate": {
          "nulls": 0,
          "distinct": 10,
          "min": 0.0062,
          "max": 0.0952
        },
        "batch_size": {
          "nulls": 0,
          "distinct": 10,
          "min": 24,
          "max": 116
        },
        "dropout_rate": {
          "nulls": 0,
          "distinct": 10,
          "min": 0.25,
          "max": 0.71
        },
        "optimizer": {
          "nulls": 0,
          "distinct": 3,
          "min": "Adam",
          "max": "SGD"
        }
      }
    },
    "gpu_utilization": {
      "rows": 600,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 600,
          "min": 1,
          "max": 600
        },
        "inference_request_id": {
          "nulls": 0,
          "distinct": 352,
          "min": 2,
          "max": 500
        },
        "gpu_id": {
          "nulls": 0,
          "distinct": 600
        },
        "utilization_percentage": {
          "nulls": 0,
          "distinct": 99,
          "min": 0,
          "max": 100
        },
        "timestamp": {
          "nulls": 0,
          "distinct": 600,
          "min": "2023-01-01 18:40:11",
          "max": "2023-12-31 23:43:10"
        }
      }
    },
    "performance_metrics": {
      "rows": 100,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 100,
          "min": 1,
          "max": 100
        },
        "model_id": {
          "nulls": 0,
          "distinct": 29,
          "min": 1,
          "max": 30
        },
        "accuracy": {
          "nulls": 0,
          "distinct": 99,
          "min": 0.0066,
          "max": 1
        },
        "latency": {
          "nulls": 0,
          "distinct": 92,
          "min": 1.47,
          "max": 9.59
        },
        "f1_score": {
          "nulls": 0,
          "distinct": 99,
          "min": 0.0364,
          "max": 0.9999
        }
      }
    },
    "user_interactions": {
      "rows": 200,
      "columns": {
        "id": {
          "nulls": 0,
          "distinct": 200,
          "min": 1,
          "max": 200
        },
        "inference_request_id": {
          "nulls": 0,
          "distinct": 163,
          "min": 1,
          "max": 498
        },
        "user_id": {
          "nulls": 0,
          "distinct": 200
        },
        "interaction_time": {
          "nulls": 0,
          "distinct": 200,
          "min": "2023-01-01 15:35:42",
          "max": "2023-12-31 11:05:05"
        },
        "feedback": {
          "nulls": 0,
          "distinct": 199
        }
      }
    }
  },
  "artifacts": [
    {
      "file": "csv/hyperparameters.csv",
      "format": "csv",
      "table": "hyperparameters",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/hyperparameters.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/hyperparameters.csv",
      "bytes": 285,
      "sha256": "13ed5c05350fe31ff1d68a340550f55aa375890a66b07535fa30b08a6efcd36c",
      "rows": 10,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "csv/models.csv",
      "format": "csv",
      "table": "models",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/models.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/models.csv",
      "bytes": 1004,
      "sha256": "8c7bc3b4e33d6f3ad97670f0a6ce7ec231a119fd1f463b22d38458d891476565",
      "rows": 30,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "csv/inference_requests.csv",
      "format": "csv",
      "table": "inference_requests",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/inference_requests.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/inference_requests.csv",
      "bytes": 21269,
      "sha256": "0f19fc907411d5dd34029fd1588950d84825a24bcbf8e153d654fde70c59e1c3",
      "rows": 500,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "csv/gpu_utilization.csv",
      "format": "csv",
      "table": "gpu_utilization",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/gpu_utilization.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/gpu_utilization.csv",
      "bytes": 41213,
      "sha256": "f0ff04c7be9da8b953c057cf586c9debe18a66018804b7ba5eb5ea874f8418a4",
      "rows": 600,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "csv/performance_metrics.csv",
      "format": "csv",
      "table": "performance_metrics",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/performance_metrics.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/performance_metrics.csv",
      "bytes": 2602,
      "sha256": "7325d093664902f2db7259c9d039cc3c070a37b98ba8428d24184fee92c4cadd",
      "rows": 100,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "csv/user_interactions.csv",
      "format": "csv",
      "table": "user_interactions",
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/user_interactions.csv",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/csv/user_interactions.csv",
      "bytes": 24320,
      "sha256": "c91962540a07f80b90013d8f78fcb5041ae318d73f12e8cd4dfba0133a26a483",
      "rows": 200,
      "contentType": "text/csv; charset=utf-8"
    },
    {
      "file": "database-script-for-high-performance-neural-network-inference-framework.json",
      "format": "json",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/database-script-for-high-performance-neural-network-inference-framework.json",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/database-script-for-high-performance-neural-network-inference-framework.json",
      "bytes": 275348,
      "sha256": "ba8258baf88a5ec9a23dac2763aa60eaadd27784473b5351c802d4828c9fcd2a",
      "rows": null,
      "contentType": "application/json; charset=utf-8"
    },
    {
      "file": "sql/postgres.sql",
      "format": "sql-postgres",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/postgres.sql",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/postgres.sql",
      "bytes": 114361,
      "sha256": "b9a0eb38745e222ff17ff4da5d88910449230a10e2634342d379788ef8dcda62",
      "rows": null,
      "contentType": "application/sql; charset=utf-8"
    },
    {
      "file": "sql/mysql.sql",
      "format": "sql-mysql",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/mysql.sql",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/mysql.sql",
      "bytes": 112082,
      "sha256": "ab5b8c66341a45dc08b56d87e7d4ee92d84f1ad70a410a339b2bdd01e7d8f03d",
      "rows": null,
      "contentType": "application/sql; charset=utf-8"
    },
    {
      "file": "sql/mssql.sql",
      "format": "sql-mssql",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/mssql.sql",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/mssql.sql",
      "bytes": 114925,
      "sha256": "51abc0faacb43f2e25f941795a26c324cc1a45c9b6682873c05ee476c252eacb",
      "rows": null,
      "contentType": "application/sql; charset=utf-8"
    },
    {
      "file": "sql/sqlite.sql",
      "format": "sql-sqlite",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/sqlite.sql",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/sql/sqlite.sql",
      "bytes": 111795,
      "sha256": "2b4eeaef17d95347a7d5598cbca9c04e19debf88c12397455cc5e7bee8a12a07",
      "rows": null,
      "contentType": "application/sql; charset=utf-8"
    },
    {
      "file": "prisma/seed.ts",
      "format": "prisma-seed",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/prisma/seed.ts",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/prisma/seed.ts",
      "bytes": 236687,
      "sha256": "38b43a2f905878f8dcb1d3a25a9a0078a0b7740a7b9a79c354d3bc663d4d385a",
      "rows": null,
      "contentType": "application/typescript; charset=utf-8"
    },
    {
      "file": "django/fixtures.json",
      "format": "django-fixture",
      "table": null,
      "key": "datasets/database-script-for-high-performance-neural-network-inference-framework/v3/django/fixtures.json",
      "url": "https://database-sample-website.s3.eu-north-1.amazonaws.com/datasets/database-script-for-high-performance-neural-network-inference-framework/v3/django/fixtures.json",
      "bytes": 348612,
      "sha256": "2a55c785420a87478cacd97c0f973c74c8fcc78c30c7e5368395a2e57c9d1ef0",
      "rows": null,
      "contentType": "application/json; charset=utf-8"
    }
  ],
  "publishedAt": 1788766227679,
  "updatedAt": 1788783355640
}