| Company database management system based on Django | Feb 2020 - Apr 2020 |
|---|---|
|
1. Designed ER model and connected front-end (HTML + CSS), back-end
and database (postgre) based on Django frame. 2. Conducted operations: insert, delete, update, search, error checking and data aggregation in Django views and forms. 3. Designed login, register, welcome page and tables-related interfaces and realized functions including dropdown menu, timepicker, interface jump, search box, tables, button and text input in Django template. |
|
| Virtual human Android system based on sentiment and voice interaction | Apr 2017 - Jun 2018 |
|
1. Accessed TURING Robot API in Android Studio to realize man-machine
conversation. 2. Processed natural language of users with jieba to realize word segmentation, obtain sentiment words and degree words. 3. Conducted weighted calculation of sentiment words and degree words based on Semantic Orientation Pointwise Mutual Information (SO-PMI) algorithm. nterface jump, search box, tables, button and text input in Django template. |
|
| Predict post-surgery risk of patients | Feb - Mar 2020 |
|
1. Data preparation: deleted duplicate value and dropped columns with
large null values. Filled null numeric data with mean value and object
data with pad method. Transformed object type into numeric type via
one-hot and LabelEncoder. Scaled data and draw correlation maps to
drop highly related data. Split data into train, test and validation
set. 2. Modeling: Clustered the target using k-means. Built Decision Tree, Multinomial Logistic Regression, LightGBM, XGBoost and CatBoost model to predict risk. Improved accuracy and prevented overfit with L1 regularization and hyper tuning tools (hyperband). 3. Evaluation: calculated AUC and Logloss. Chose best model to predict new data. |
|
| Musical Instruments | Games | Sports |
|---|---|---|
| Guitar, Guzheng, Kalimba | Overcook, Human Fall Flat, Fall Guys | Hiking |