Qiang Daijun

• 2023Developed a deep learning model for accurate facial emotion classification.
• - Curated and preprocessed a diverse dataset of labeled facial expressions.
• - Designed a Convolutional Neural Network (CNN) with advanced techniques for improved model performance.
• - Achieved a testing dataset accuracy of 84%, surpassing random guessing.
• - Selected the most effective model for further development.
• - Demonstrated the potential of deep learning in accurate facial emotion classification, with broad applications in fields such as security.

• Conducted teaching tutorials for an undergraduate class of 20 students, prepared lessons and created teaching materials and practice problems based on the syllabus - Lecture content: R usage, R statistical visualization methods, statistical models, data exploration, analysis of variance, regression analysis, sample probabilities respectively, etc

• Tencent News Search Text Analysis and Modeling - Explored millions of user search data by python through word frequency and sentiment analysis, plotted word clouds and time series trend for each key word - Analyzed the distribution of primary and secondary categories of Tencent news postings by Tableau; Trained a classifier model using Naïve Bayes and deep learning methods to predict the number of posts in selected time windows by R
• Cancer Patient Case Data Analysis - Used LightGBM and gradient learning framework to analyze the incidence data of cases and patients - Used deep learning for data mining and built a model with accuracy rate of 90%
• Data Processing of e-commerce Home Appliance Products - Applied Python for comment word cloud, research on GMV distribution, users’ scores for products, correlation test for models that clients purchased, analysis of data variation - Drew the autocorrelogram for GMV, users’ reviews and scores and brand, used Tableau to visualize the correlation between brand and GMV, users’ reviews and scores; composed the PPT and made presentation e-Commerce Data Analysis - Cleaned the given data through classification, screening and reorganization - Defined and calculated metrics such as page views, clicks ,exposure rate - Analyzed and visualized the correlations between the features and consumer’s shopping decisions, presented the results to the team and leaders with PPT

Introduction: the project conducted model training for the number of people and time of day in New York restaurants to predict the average number of meal arrivals after 10 days

- Divided the data set in the last 6 months to training set and test set, and used ARIMA model to find out the autocorrelation of historical data

- Used logarithm to eliminate the data fluctuation and used the first order difference to eliminate the trend growth and ensure the sequence stability

- Performed the white noise verification of the data, calculated ACF, PACF and used ARIMA for recognition; used ARIMAX() to input the parameters of the model; drew the ACF and PACF graphs of data, and studied the lagged variable

- Adopted the state space in StatsModels for fitting after the determination of parameters, and obtained the result which was very close to parameters (90%) to determine the feasibility of the project

- Collected COVID-19 stats data and understood the medical situation in Mumbai in response to the 2020 epidemic from online resources (NSS,ICRC, WHO)

- Conducted budget planning(costofnucleicacid,vaccination,quarantine,isolationequipment,etc.) and visualized population mobility using python

- Coordinated the communication among team members; organized brainstorming to ensure the successful completion of the project

- Proposed comprehensive prevention such as keeping social distance, setting up inspections in surrounding areas, and isolating infected communities, etc