LTF Farmer Income Prediction Challenge

Eligibility: The competition is open to college/university students. Participants may register individually or as a team, subject to the team-size requirements 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 21 September, 9:00 AM: Project Submission Deadline 22 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.