• Given labeled data sets from a car insurance company, performed EDA, data preprocessing and feature engineering
• Trained an ensemble classification model to predict fraud of insurance in the test data set, resulting in an 86?curacy

Pre-processed the dataset by visualization & feature engineering; selected the most correlated features for machine learning Trained an XGBoost model on the processed data; performed prediction on the test set, which resulted in being top 11%

• Preprocessed & visualized the US. Business & Industry data sets; conducted decomposition & ACF analysis on the time series
• Selected an ARIMA model on the training set by comparing the AIC of different parameters, and achieved a WAPE of 3.5%