{
  "fetched_at": "2026-10-19T21:01:50Z",
  "source_url": "https://www.kaggle.com/competitions/gears-fault-detection-through-lubrication/data",
  "pages": [
    {
      "name": "data-description",
      "content": "#**Variables:**\n\n##**1. Temperature (Temp):**\n\n- **Description:** Temperature of the gearbox lubricant.\n- **Type:** Numeric (float).\n- **Units:** Degrees Celsius.\n\n##**2. Viscosity (Visc):**\n\n- **Description:** Viscosity of the lubricant, a measure of its resistance to flow.\n- **Type:** Numeric (float).\n- **Units:** Centistokes (cSt).\n\n##**3. Pressure (Pres):**\n\n- **Description:** Pressure within the gearbox.\n- **Type:** Numeric (float).\n- **Units:** Pascals (Pa).\n\n##**4. Metal Particle Content (Metal):**\n\n- **Description:** Concentration of metal particles in the lubricant.\n- **Type:** Numeric (float).\n- **Units:** Parts per million (ppm).\n\n##**5. Water Content (Water):**\n\n- **Description:** Percentage of water content in the lubricant.\n- **Type:** Numeric (float).\n- **Units:** Percentage (%).\n\n##**6. Additive Levels (Additive):**\n\n- **Description:** Levels of additives present in the lubricant (e.g., anti-wear additives).\n- **Type:** Numeric (float).\n- **Units:** Parts per million (ppm).\n\n##**7. pH:**\n\n- **Description:** pH value of the lubricant.\n- **Type:** Numeric (float).\n- **Units:** pH scale (0-14).\n\n##**8. Density (Dens):**\n\n- **Description:** Density of the lubricant.\n- **Type:** Numeric (float).\n- **Units:** Grams per cubic centimeter (g/cm³).\n\n##**9. Conductivity (Cond):**\n\n- **Description:** Electrical conductivity of the lubricant.\n- **Type:** Numeric (float).\n- **Units:** Siemens per meter (S/m).\n\n##**10. Oxidation Index (Oxid):**\n\n- **Description:** Oxidation stability index of the lubricant.\n- **Type:**Numeric (float).\n- **Units:** Arbitrary units (AU).\n\n#**Target Variable:**\n\n##**Gearbox State (Target):**\n**Description:** Binary variable indicating the state of the gearbox.\n\n**Values:**\n- 0: Good condition.\n- 1: Faulty condition.\n\n#**Dataset Details:**\n\n***Training Data:*** Contains labeled examples of gearbox states based on lubrication data for model training.\n\n***Test Data:*** Unlabeled dataset used for final evaluation, where participants submit predictions for gearbox states.\n\n***Note:*** Ensure that missing values (NaNs) and outliers are properly handled before training your models to ensure robust performance.\n\n##**Files**\n\n*   **train.csv** - the training set\n*   **test.csv** - the test set\n*   **sample_submission.csv** - a sample submission file in the correct format\n*   **metaData.csv** - supplemental information about the data\n\n## Columns\n\n*   `example_id` - definition of example_id\n*   `feature_1` - definition of feature_1\n*   etc.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Description",
      "content": "Gearboxes are critical components in industrial machinery, responsible for transmitting mechanical power through a system of gears. Predicting faults in gearboxes is essential for preventing costly breakdowns and optimizing maintenance schedules. In this competition, participants are challenged to develop machine learning models that can accurately classify the condition of gearboxes based on data collected from their lubrication systems.\n\n**Task:**\n\nParticipants will work with a dataset containing various multivariate features extracted from gearbox lubrication processes. These features include temperature, viscosity, pressure, metal particle content, water content, additive levels, pH, density, conductivity, and oxidation index. The objective is to build a classification model that can predict whether a gearbox is in a normal operating condition or if it exhibits signs of potential faults.\n\n**Data Description:**\n\nTraining Data: The training dataset consists of labeled examples where each instance includes measurements of the aforementioned variables along with a label indicating the gearbox's state (Good or Faulty).\n\nTest Data: The test dataset contains instances with the same set of variables but without the corresponding labels. Participants are required to submit predictions for these instances, indicating the likelihood of a gearbox fault.\n\n**Evaluation Metric:**\n\nSubmissions will be evaluated based on their classification accuracy using metrics such as precision, recall, and F1-score. The primary goal is to correctly identify gearboxes with faults while minimizing false positives and false negatives.\n\n**Submission Format:**\n\nParticipants are required to submit a CSV file with predictions for each instance in the test set. Each prediction should indicate the probability or confidence score of the gearbox being faulty.\n\n**Prizes:**\n\nExciting prizes await the top-performing participants based on their leaderboard rankings. Additionally, participants will gain recognition within the Kaggle community for their innovative solutions to gearbox fault detection.\n\n**Join Us:**\n\nGear up and join this competition to showcase your machine learning expertise in predicting gearbox faults based on industrial lubrication data. Together, we can revolutionize gearbox maintenance practices and improve operational reliability across industries.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "Evaluation",
      "content": "**Evaluation Process: Accuracy**\n\nAccuracy is a fundamental metric used to evaluate classification models in machine learning. It measures the ratio of correct predictions to the total number of predictions made by the model.\n\n**Step-by-Step Evaluation using Accuracy:**\n\n**True Positives (TP):**\n- Number of correctly predicted instances where the gearbox is faulty.\n\n**True Negatives (TN):**\n- Number of correctly predicted instances where the gearbox is in good condition.\n\n**False Positives (FP):**\n- Number of incorrectly predicted instances where the model predicted the gearbox is faulty, but it is actually in good condition.\n\n**False Negatives (FN):**\n- Number of incorrectly predicted instances where the model predicted the gearbox is in good condition, but it is actually faulty.\n\n**Confusion Matrix:**\n- Before calculating accuracy, a confusion matrix is constructed from the predictions and the true labels of the test dataset:\n\n|                | Predicted: Good      | Predicted: Faulty    |\n|----------------|----------------------|----------------------|\n| Actual: Good   | True Negative (TN)   | False Positive (FP)  |\n| Actual: Faulty | False Negative (FN)  | True Positive (TP)   |\n\n\n**Accuracy Calculation:**\n\n- Accuracy is calculated as the ratio of correct predictions (both true positives and true negatives) to the total number of predictions:\n\nAccuracy = (TP+TN) \\ (FP+FN+TP+TN​)\n\n- It represents the overall correctness of the model's predictions.\n\n**Mean Accuracy:**\n\nFor competitions with multiple classes or instances, the mean accuracy is often used to average the accuracies across all classes or instances. This provides a comprehensive measure of the model's overall performance.\nSubmission Requirement:\n\nParticipants are required to submit predictions in the form of probabilities or confidence scores for each instance in the test dataset.\nKaggle's evaluation system will calculate the accuracy based on these predictions against the true labels of the test dataset.\nThe leaderboard ranks participants based on their accuracy scores, with higher scores indicating better-performing models in predicting gearbox faults.\n\n**Conclusión:**\n\nAccuracy is a straightforward and widely used metric for evaluating classification models in this competition. It measures the model's ability to correctly classify gearboxes as faulty or in good condition based on the provided data.\nIt is essential for participants to aim for high accuracy scores to demonstrate the effectiveness of their models in gearbox fault detection.\n\n## Submission File\nFor each ID in the test set, you must predict a probability for the Prediction variable. The file should contain a header and have the following format:\n\n    ID,Prediction\n    2,Good\n    3,Good\n    4,Bad\n    etc.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "rules",
      "content": "Certainly! Here are the rules for your Kaggle competition:\n\n---\n\n**Competition Rules: Gearbox Fault Detection Challenge**\n\n1. **Objective:**\n   - Participants must develop machine learning models to predict gearbox faults based on industrial lubrication data.\n\n2. **Dataset:**\n   - Provided datasets include multivariate features such as temperature, viscosity, pressure, metal particle content, water content, additive levels, pH, density, conductivity, and oxidation index.\n   - Training data: Labeled examples of gearbox states (good or faulty).\n   - Test data: Unlabeled dataset for final evaluation.\n\n3. **Evaluation:**\n   - Submissions are evaluated based on the accuracy metrics.\n   - Higher scores indicate better-performing models in predicting gearbox faults.\n\n4. **Submission Format:**\n   - Participants must submit predictions for the test dataset indicating the probability or confidence of gearbox faults.\n\n5. **Team Composition:**\n   - Teams are allowed with a máximum of 6 participants per team.\n\n6. **Code and Documentation:**\n   - Participants must provide well-documented code that includes data preprocessing, model training, and evaluation.\n   - Models should be reproducible and efficient in predicting gearbox faults.\n\n7. **Fair Play:**\n   - All submissions must adhere to Kaggle's code of conduct and competition rules.\n   - Plagiarism and unauthorized use of external datasets are strictly prohibited.\n\n8. **Prizes:**\n   - Top-performing participants will receive prizes based on their leaderboard rankings.\n   - Prizes are awarded at the discretion of the competition organizers.\n\n9. **Updates and Communication:**\n   - Organizers may provide updates, clarifications, or changes to the competition rules through Kaggle forums or announcements.\n   - Participants are encouraged to engage actively and respectfully in discussions related to the competition.\n\n10. **Final Decision:**\n    - The final decision on winners and competition outcomes rests with the competition organizers.\n    - Decisions are based on the evaluation metrics and compliance with competition rules.\n\n11. **Disqualification:**\n    - Participants may be disqualified for violating competition rules or engaging in misconduct.\n    - Organizers reserve the right to disqualify participants at their discretion.\n\n12. **Disclaimer:**\n    - Organizers are not responsible for any technical issues, data discrepancies, or disruptions during the competition.\n    - Participation in the competition implies acceptance of all rules and conditions stated herein.\n\n---\n\nThis is a page where you can include rules that participants must accept before joining. You may wish to include rules like:\n\n - Don't cheat!\n - Apply yourself!\n - Have fun!",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    },
    {
      "name": "abstract",
      "content": "Welcome to the Gearbox Fault Detection Challenge!\n\nIn this competition, participants are tasked with developing a robust classification model to predict faults in gearboxes based on industrial lubrication data. Gearboxes are crucial components in various mechanical systems, and predicting faults early can significantly enhance maintenance practices and prevent costly breakdowns.",
      "isPublished": false,
      "hasIsPublished": false,
      "postTitle": "",
      "hasPostTitle": false,
      "mimeType": "",
      "hasMimeType": false
    }
  ]
}