This is a Course Project for Data Mining(DATA620007) in School of Data Science, Fudan University.
- Tensorflow 1.0.1
- Python 3.6
Crawled from Centanet at April 2018.
The crawling part was constructed by Yan Zhiquan.
Structure: Two-hidden-layer Full Connect Neural Network with 200 nodes and ReLU for each, as well as a Dropout with keep ratio of 75%.
Epoch: 10000
MSE(on testing set): < 0.03
A simple example can be seen as follows, witch means the fair price of the dormitory we lived may worth the value up to ¥5.8 million.
The output includes the average price and the total price of the target real easte.