Club Mahindra (Club M) makes significant revenue from Food and Beverages (F&B) sales in their resorts. The members of Club M are offered a wide variety of items in either buffet or À la carte form. Following are some benefits that the model to predict the spend by a member in their next visit to a resort will bring:
- Predicting the F&B spend of a member in a resort would help in improving the pre-sales during resort booking through web and mobile app.
- Targeted campaigns to suit the member taste and preference of F&B
- Providing members in the resort with a customized experience and offers
- Help resort kitchen to plan the inventory and food quantity to be prepared in advance
Given the information related to resort, club member, reservation etc. the task is to predict average spend per room night on food and beverages for the each reservation in the test set.
train.zip contains train.csv and data_dictionary.csv.
- train.csv contains the training data with details on a set of reservations with the average spend per room night
- Data_Dictionary.xlsx contains a brief description of each variable provided in the training and test set.
test.csv contains details of all reservations for which the participants need to predict the average spend on FnB per room night
sample_submission.csv contains the submission format for the predictions against the test set. A single csv/zip needs to be submitted as a solution.
Submissions are evaluated on 100 * Root Mean Squared Error (RMSE) on the variable amount_spent_per_room_night_scaled
To qualify for the Prizes, the RMSE of your submitted predictions on the private set or the private leaderboard score must beat the baseline RMSE set by Club Mahindra = 97
Test data is further randomly divided into Public (30%) and Private (70%) data. * Your initial responses will be checked and scored on the Public data. * The final rankings would be based on your private score which will be published once the competition is over.
Private LB Score of 97.2439527785 78th/983 || Public LB sscore 96.2403981833 84th/983
- I havent updated the hyper parameter of LGB , XGB and CatBooster as seen in other solution
- I have seen one solution where they have used more tree depth (Which i tried but i tought its overfitting), i will update the solution with hyper parameters
- More Investigation into the Data