LTF Farmer Income Prediction Challenge

Join the Whatsapp Community: https://chat.whatsapp.com/GD1WH6iq43qIGR7QwVtI7F

Eligibility:

• The competition is open to college/university students.
• Participants may register only as a team, subject to the team-size requirements (2-5) specified on the competition portal.
• Inter-college teams are permitted, unless otherwise specified by the organizers.
• Inter-specialization/branch teams are permitted. Participants from engineering, management, finance, data science, economics, statistics, and other relevant disciplines are encouraged to participate.
• Each participant must provide valid institutional details during registration.

Challenge Format:

• The competition is based on a real-world case study provided by L&T Finance (LTF).
• Participants will be provided with:
• Training Dataset: Containing farmer demographic, landholding, climatic, and living-index indicators.
• Test Dataset: For generating income predictions.
• Data Dictionary: Explaining the variables provided.
• The primary task is to develop a machine learning/data-driven model to predict farmer income using the provided dataset.
• Participants may explore and incorporate relevant publicly available external datasets or data sources to improve their analysis and predictions, wherever appropriate.
• The evaluation metric for the prediction task will be Mean Absolute Percentage Error (MAPE).
• Participants are expected to submit their working Python code along with the prediction output in the prescribed format.
• The approach/methodology document will be required only for teams shortlisted for the final presentation round.

Submission:

• The prediction file must strictly follow the prescribed submission format: TeamName_CollegeName_IdentityNumber.csv
• The output file must conform to the sample format provided on the competition portal.
• Submissions must be made before the stipulated deadline. Late submissions may not be considered.
• Teams shortlisted for the final round will be required to present their approach and findings to the L&T Finance panel.

Timeline:

• 17 September: Contest Launch
• 22 September, 9:00 AM: Project Submission Deadline
• 23 September: Top 8 Teams Announced
• 26 September: Final Presentations and Felicitation at NITK
• Each shortlisted team will receive a 15-minute presentation slot.

Rules:

• Participants must submit original work. Plagiarism, copying, or unauthorized use of another team's work is strictly prohibited.
• Participants may use publicly available datasets, research papers, blogs, tutorials, open-source libraries, and other resources, provided they are used appropriately and properly acknowledged/cited where applicable.
• Use of external datasets or features is permitted, provided their source and methodology can be explained during the presentation.
• Participants must not manipulate, fabricate, or deliberately misrepresent data or results.
• The submitted prediction file must strictly adhere to the prescribed format and contain the required fields.
• Only submissions received within the specified deadline will be considered for evaluation.
• Participants must be able to explain their data preprocessing, feature engineering, modelling approach, validation strategy, and results if shortlisted.
• Shortlisted teams must be available for the final presentation on 26 September.
• L&T Finance and the organizing team reserve the right to disqualify submissions involving plagiarism, fraudulent data, violation of competition rules, or other forms of misconduct.
• In case of any dispute regarding the competition, the decision of the organizers and L&T Finance shall be considered final.
• By participating, teams agree to the use of their submitted work for evaluation and competition-related purposes.

Evaluation:

The competition will primarily assess:

• Quality of the problem-solving approach.
• Data understanding and preprocessing.
• Feature engineering and modelling methodology.
• Model performance based on MAPE.
• Ability to derive meaningful and actionable insights.
• Clarity and robustness of the proposed solution.

Note: While model accuracy is an important component, the competition places significant emphasis on the thought process, methodology, and quality of insights demonstrated by the participants.