Forecasting Sticker Sales:
This repository contains the code and models used for the Kaggle competition 'Forecasting Sticker Sales'. The objective was to predict future sticker sales based on historical data. The dataset includes over 200,000 rows of data. I used Random Forest Regressor and LGBM Regressor for modeling and applied hyperparameter tuning techniques to achieve the best possible performance.
Models Used:
Random Forest Regressor: A robust machine learning algorithm used for regression tasks. Tuned using hyperparameters to improve accuracy. LGBM Regressor: A gradient boosting method designed for speed and performance. It was also fine-tuned using hyperparameters for optimal performance. I used hyperparameter tuning technique Grid Search for both models
Kaggle Leaderboard:
my model ranks 1090th out of 2722 participants on the Kaggle leaderboard. Iam Happy that iam improving my self compare to previous competition
Data:
The dataset used for this competition contains 200,000+ rows of historical sticker sales data along with customer demographic and other relevant features. The data was cleaned and preprocessed before training the models.