Haowen Rui

Experience

• 

  Fondo Atlas

  Jan 2022 - Mar 2022

  Data Science Consultant

  ● Performed exploratory data analysis and data cleaning with Python on a dataset including every multi-family property in the US market to investigate factors affecting rent per square feet and identify potential undervalued properties.● Researched nationwide location data for Starbucks, Wholefoods, and hospitals and calculated the number of such amenities within 1 mile of each property through longitude and altitude in Python. Used Tableau to visualize the relationship between rent per square feet and all possible features.● Used the Census Tract API and assigned Census Tract Code for each property.● Applied 4 different machine learning models (multiple regression, decision tree, random forest, and neural network) and compared their performance based on Mean Squared Error. Random Forest turned out to have the best performance.● Selected 6 properties with the largest potential return based on the model results, collected information about their surroundings and environments, and recommended them to the company for future investment. Show less

• 

  Uber (Hong Kong), Ltd.

  Aug 2022 - Oct 2022

  Business Analyst

  ● Performed exploratory data analysis and data cleaning with Python on a dataset of ride-hailing orders to investigate factors affecting whether a customer will cancel a placed order.● Categorized useful numerical features to help with classification. Performed Log and Square Root transformation on severely right skewed features, and trained a logistic regression model using Scikit Learn.● Revised each variable’s statistical significance using ‘statsmodels’ package. Found out that whether a user has membership does not show statistical significance on predicting their probability of canceling order.● Reran the model after removing insignificant variables. Finalized model had 84.683% prediction accuracy on testing data, with an ROC score of 89.763% and all variables being statistically significant.● Based on statistical significance summary, concluded that whether a customer used discount code has a significant impact on order canceling likelihood. Customers who used a discount code are more likely to cancel the ride.● Presented the results to the manager and suggested to shorten the duration of time for free cancellation or increase the cancellation fee for riders who used coupons to reduce possibility of cancellation and optimize expected profit. Show less

• 

  Schneider Medical Industries

  Oct 2023 - now

  Jr. Product and Marketing Analyst