ML Competition

Description

The ML Competition is an online individual machine-learning competition focused on predictive modelling, data analysis and applied machine-learning problem solving. Participants will work with the provided problem statement and dataset to understand the data, preprocess inputs, develop suitable machine-learning models, evaluate their approaches and generate the required predictions.

Participants must follow the specified dataset, tools, evaluation metric, submission format and competition rules. Submissions will be evaluated using the published performance metric and may undergo technical verification before the final results are announced.

Duration

- Total Duration: 90–120 minutes

Rules & Guidelines

• Eligibility
• Participation is individual.
• Participants must satisfy the eligibility requirements specified in the official event announcement.
• Participants must provide accurate registration information.
• Each participant may submit only within the limits specified by the competition rules.
• Competition Format
• Participants will receive a machine-learning problem statement and dataset.
• Participants must analyze the supplied data and develop an appropriate predictive model.
• The competition may include training and test data, as specified by the organizers.
• Final predictions must be submitted in the required format.
• Dataset Usage
• Participants must use the dataset and data sources permitted by the organizers.
• Participants must not manipulate or access hidden test labels.
• Data leakage, unauthorized access to evaluation data or attempts to obtain hidden answers are prohibited.
• Any additional datasets or external data must comply with the published competition rules.
• Tools & Technologies

Participants may use the Python/ML libraries and tools explicitly permitted by the organizers.

Examples may include:

• Python
• NumPy
• Pandas
• Scikit-learn
• Matplotlib/Seaborn
• Approved machine-learning frameworks

The exact supported tools and versions should be published before the competition begins.

5. Model Development

Participants may perform appropriate:

• Data preprocessing
• Exploratory data analysis
• Feature engineering
• Feature selection
• Model training
• Model validation
• Hyperparameter tuning
• Prediction generation

All techniques must comply with the published competition rules.

6. Submission

Participants must submit the required prediction file and any additional code or documentation specified by the organizers.

The submission must:

• Follow the required file format.
• Contain the required columns/fields.
• Be submitted through the designated platform.
• Be submitted before the official deadline.
• Evaluation & Scoring

Submissions will be evaluated using the official performance metric announced for the competition.

Depending on the challenge, this may include metrics such as:

• Accuracy
• Precision
• Recall
• F1-score
• Mean Absolute Error
• Mean Squared Error
• Root Mean Squared Error
• ROC-AUC

The exact metric must be specified in the final competition rules.

8. Reproducibility & Verification

Organizers may review high-ranking or unusual submissions to verify:

• Reproducibility
• Data leakage
• Compliance with the dataset rules
• Use of permitted resources
• Validity of the submitted solution

Participants may be asked to provide code or supporting documentation where required.

9. External Resources & AI Tools

The use of external datasets, pretrained models, APIs, online resources and AI-assisted tools will be governed by the official competition policy.

Participants must comply with any restrictions or disclosure requirements.

• Originality
• Participants must independently develop their solutions unless collaboration is explicitly permitted.
• Sharing or copying another participant's solution is prohibited.
• Code similarity or integrity checks may be performed.
• Violations may result in disqualification.
• Disqualification

A participant may be disqualified for:

• Ineligible participation
• Data leakage
• Accessing hidden test information
• Manipulating the evaluation system
• Using prohibited external data/resources
• Copying another participant's solution
• Unauthorized collaboration
• Submitting after the deadline
• Providing false information
• Violating the published competition rules
• Tie-Breaking
• If participants achieve the same final score, the published tie-break mechanism will be applied consistently.
• Final Results
• The leaderboard/results will be finalized after the required technical and eligibility verification.