Beijing Housing Price Prediction (Python, Vue.js, Django)
- Used scikit-learn framework to do simulated training based on optimized
Classiﬁcation and Regression Tree, combined it with the K nearest neighbors
algorithm and got the optimal housing model.
- Showed data virtualization by Pandas, NumPy, matplotlib, and displayed the performance of decision tree.
- Input 10,000 groups data of housing price as a practice, and achieved 78.92% model performance.
Facial Attractive Computing based on CNN (Python, Django)
- Used TensorFlow framework to build CNN network, combined such technologies
as face alignment, multiregional feature extraction of human faces etc.
- Got the alignment and redundant feature through facial landmark detection and
rotation, and multiregional feature extraction of human faces, thus enhanced CNN performance.
Qingdao University - Mobile APP Backend (Python, Django, RESTful)
- Used Python Django to implement Campus Lost and Found, in-campus navigation and score inquiry, etc.
Sensor Data Distributed Storage System (C++, Java)
- Implemented interface between sensors and Raspberry Pi and real-time sending
sensor data to storage layer.
- Enhanced the efficiency of data writing, used Hadoop Cluster, combined them
with a temporary storage layer based on Redis between the gateway and storage
layer, and thus saved 30% of time during data writing.
Raspberry Online Barrage (Node.js, FFmpeg)
- Implemented the basic barrage function of bilibili.com
- Using Raspberry Pi as server, support for up to 50 clients to watch the live.