NUS Datathon Champion Team L

What it does

The challenge presented is to develop a robust predictive model capable of forecasting future sales figures for these companies. The objective is to mine the dataset to identify trends, patterns, and potential causative factors that significantly impact sales outcomes. By integrating historical sales data with predictive analytics, the goal is to achieve accurate sales forecasts that companies can utilize to make informed strategic decisions, optimize inventory management, and allocate resources effectively.

How we built it

Using feature selection methods, we cleaned the dataset and selected relevant features. Then, we represented categorical data using numerical data via one-hot encoding and label encoding, then handled missing data using KNN Imputation. Next, we created a correlation matrix to observe the relationships between different features and represented them through visualisation methods. We explored different models such as linear regression, decision tree regressor, random forest regressor and then the XGBoost model to decide on the most appropriate model.

Built With

• matplot
• pandas
• python
• seaborn
• sklearn