Skip to content
Predict first day performance of Hong Kong IPO stocks: A pipeline example of machine learning projects
Branch: master
Clone or download
Latest commit e0c512d Jul 22, 2017
Permalink
Type Name Latest commit message Commit time
Failed to load latest commit information.
aastock_crawler first commit Jul 20, 2017
data add data Jul 20, 2017
feature_engineering fix plotting bug Jul 21, 2017
img fix plotting bug Jul 21, 2017
machine_learning fix plotting bug Jul 21, 2017
.gitignore add data Jul 20, 2017
README.md add blog Jul 22, 2017

README.md

hk_ipo_prediction

中文blog

Predict first day performance of Hong Kong IPO stocks: A pipeline example of machine learning projects.

Procedures:

  1. Crawl, Parse and Clean Hong Kong IPO data from AAStocks.com using selenium webdriver and phantomjs (around 400 data points).

  2. Use pandas for data cleaning and feature engineering, including feature selection and handling big values, missing values and categorical values (one hot encoding)

  3. Use xgboost for regression model to predict first day performance. Generated feature importance plot is very interesting.

Feature Importance Plot

You can’t perform that action at this time.