TBrain - E.SUN House Price Prediction

4th place solution to the TBrain competition - E.SUN AI Open Competition Summer 2019 - House Price Prediction (玉山人工智慧公開挑戰賽2019夏季賽 - 台灣不動產AI神預測)

Contributors

Chen-Hsi (Sky) Huang (github.com/skyhuang1208)
Louis Yang (github.com/louis925)

Achievement

We won 4th place out of 766 teams (top 0.5%) with private leaderboard score 6210.877.

Documentation

Structure

dataset - place input dataset train.csv and test.csv
- gen_5_fold_cv.ipynb - generate 5-fold CV dataset
eda-and-exp/ - contains notebooks for exploratory data analysis and various experiments
- eda* - exploratory data analysis
- exp* - experiments
model-<model_number>-build-<model_name>.ipynb/.py - parameters search with small training step for single model
model-<model_number>-predict-<model_name>.ipynb/.py - complete single model training process and prediction. Output single model CV and test set prediction for stacking
stack_<stack_method>_<stacking_model_number>_<model_numbers_used>.ipynb - stacking model from single models in <model_numbers_used>. Output to final test set prediction for submission.
feature_engineering.py - label encoders and feature scalers
keras_get_best.py - early stop callback for keras
utilities.py - utility functions including scoring, feature processing, ..., etc
vars_03.py - feature importance computed by experiments for feature selection
loss_exp.ipynb - experiment on the smooth hit rate loss function
price_quantizer*.ipynb - quantize predicted price used by stack 16 and 18

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Repository files navigation

TBrain - E.SUN House Price Prediction

Contributors

Achievement

Documentation

Structure

How it work

About

Releases

Packages

Contributors 2

Languages

Name		Name	Last commit message	Last commit date
Latest commit History 6 Commits
dataset		dataset
doc		doc
eda-and-exp		eda-and-exp
README.md		README.md
blend1_search.py		blend1_search.py
blend2_combine.py		blend2_combine.py
feature_engineering.py		feature_engineering.py
keras_get_best.py		keras_get_best.py
loss_exp.ipynb		loss_exp.ipynb
metamodel-01-kmeans-latlong.ipynb		metamodel-01-kmeans-latlong.ipynb
metamodel-02-keras-autoencoder.ipynb		metamodel-02-keras-autoencoder.ipynb
metamodel-03-lgb-binary.ipynb		metamodel-03-lgb-binary.ipynb
model-01-build-lgb.ipynb		model-01-build-lgb.ipynb
model-01-predict-lgb.ipynb		model-01-predict-lgb.ipynb
model-02-build-keras-search-Copy1.ipynb		model-02-build-keras-search-Copy1.ipynb
model-02-build-keras-search-batch_size.ipynb		model-02-build-keras-search-batch_size.ipynb
model-02-build-keras-search-embedding_size.ipynb		model-02-build-keras-search-embedding_size.ipynb
model-02-build-keras-search.ipynb		model-02-build-keras-search.ipynb
model-02-predict-keras-search.ipynb		model-02-predict-keras-search.ipynb
model-03-build-lgb-corr-feats-selection.ipynb		model-03-build-lgb-corr-feats-selection.ipynb
model-03-build.py		model-03-build.py
model-03-predict-lgb-corr-feats-selection-100.ipynb		model-03-predict-lgb-corr-feats-selection-100.ipynb
model-04-build-lgb-PCA.ipynb		model-04-build-lgb-PCA.ipynb
model-04-build-lgb-PCA.py		model-04-build-lgb-PCA.py
model-04-predict-lgb-PCA.ipynb		model-04-predict-lgb-PCA.ipynb
model-05-build-lgb-wo-per-area.ipynb		model-05-build-lgb-wo-per-area.ipynb
model-05-predict-lgb-wo-per-area.ipynb		model-05-predict-lgb-wo-per-area.ipynb
model-06-build-lgb-plot_for_mae_vs_lr.ipynb		model-06-build-lgb-plot_for_mae_vs_lr.ipynb
model-06-build-lgb-search-lr-old-backup.ipynb		model-06-build-lgb-search-lr-old-backup.ipynb
model-06-build-lgb-search-lr.ipynb		model-06-build-lgb-search-lr.ipynb
model-06-predict-lgb-lr0.001.ipynb		model-06-predict-lgb-lr0.001.ipynb
model-07-build-keras-embedding.ipynb		model-07-build-keras-embedding.ipynb
model-07-predict-keras-embedding.ipynb		model-07-predict-keras-embedding.ipynb
model-08-build-keras-embedding-2.ipynb		model-08-build-keras-embedding-2.ipynb
model-08-predict-keras-search.ipynb		model-08-predict-keras-search.ipynb
model-09-predict-lgb-rm-75-no_hist.ipynb		model-09-predict-lgb-rm-75-no_hist.ipynb
model-09-predict-lgb-rm-75.ipynb		model-09-predict-lgb-rm-75.ipynb
model-10-predict-lgb-rm-75-lr-0.001.ipynb		model-10-predict-lgb-rm-75-lr-0.001.ipynb
model-11-build-rf.ipynb		model-11-build-rf.ipynb
model-11-build-rf.py		model-11-build-rf.py
model-11-predict-rf.ipynb		model-11-predict-rf.ipynb
model-12-predict-keras-search-prelu.ipynb		model-12-predict-keras-search-prelu.ipynb
model-13-predict-keras-he_uni.ipynb		model-13-predict-keras-he_uni.ipynb
model-14-predict-lgb-rm-75-lr-0.001-rand.ipynb		model-14-predict-lgb-rm-75-lr-0.001-rand.ipynb
model-15-predict-lgb-rm-75-lr-0.001-rand323.ipynb		model-15-predict-lgb-rm-75-lr-0.001-rand323.ipynb
model-16-predict-lgb-rm-68-lr-0.001-mix5.ipynb		model-16-predict-lgb-rm-68-lr-0.001-mix5.ipynb
model-17-predict-lgb-rm-70-lr-0.001-mix5.ipynb		model-17-predict-lgb-rm-70-lr-0.001-mix5.ipynb
model-18-predict-lgb-rm-70-lr-0.001-p5.ipynb		model-18-predict-lgb-rm-70-lr-0.001-p5.ipynb
model-19-build-lgb-search-bins.ipynb		model-19-build-lgb-search-bins.ipynb
model-19-predict-lgb-search-bins-lr-0.0005.ipynb		model-19-predict-lgb-search-bins-lr-0.0005.ipynb
model-20-predict-lgb-rm-70-lr-0.0008-mix5.ipynb		model-20-predict-lgb-rm-70-lr-0.0008-mix5.ipynb
model-21-predict-lgb-wo-per-area-long.ipynb		model-21-predict-lgb-wo-per-area-long.ipynb
model-22-predict-lgb-wo-per-area-long-2.ipynb		model-22-predict-lgb-wo-per-area-long-2.ipynb
model-23-build-lgb-binary.ipynb		model-23-build-lgb-binary.ipynb
model-23-predict-lgb-binary.ipynb		model-23-predict-lgb-binary.ipynb
model-24-build-lgb-binary-augment.ipynb		model-24-build-lgb-binary-augment.ipynb
model-24-predict-lgb-binary-augment.ipynb		model-24-predict-lgb-binary-augment.ipynb
model-25-predict-lgb-search-bins-lr-0.0005-250.ipynb		model-25-predict-lgb-search-bins-lr-0.0005-250.ipynb
model-26-predict-lgb-search-bins-lr-0.0005-350.ipynb		model-26-predict-lgb-search-bins-lr-0.0005-350.ipynb
model-27-predict-lgb-feat_rm_new.ipynb		model-27-predict-lgb-feat_rm_new.ipynb
model-28-predict-lgb-search-bins-lr-0.0005-255.ipynb		model-28-predict-lgb-search-bins-lr-0.0005-255.ipynb
model-29-predict-building_age-fillna.ipynb		model-29-predict-building_age-fillna.ipynb
model-30-predict-lgb-binary-2.ipynb		model-30-predict-lgb-binary-2.ipynb
model-31-predict-lgb-3_groups.ipynb		model-31-predict-lgb-3_groups.ipynb
model-32-predict-lgb-remove_outlier_01.ipynb		model-32-predict-lgb-remove_outlier_01.ipynb
model-33-predict-lgb-remove_outlier_03.ipynb		model-33-predict-lgb-remove_outlier_03.ipynb
model-34-predict-lgb-remove_outlier_01.ipynb		model-34-predict-lgb-remove_outlier_01.ipynb
model-35-predict-lgb-remove_outlier_03-0.001-kfolds.ipynb		model-35-predict-lgb-remove_outlier_03-0.001-kfolds.ipynb
model-35-predict-lgb-remove_outlier_03-0.001.ipynb		model-35-predict-lgb-remove_outlier_03-0.001.ipynb
model-36-predict-lgb-remove_outlier_03-0.001-2.ipynb		model-36-predict-lgb-remove_outlier_03-0.001-2.ipynb
model-37-predict-lgb-remove_outlier_05.ipynb		model-37-predict-lgb-remove_outlier_05.ipynb
price_quantizer_stack_16_1-27.ipynb		price_quantizer_stack_16_1-27.ipynb
price_quantizer_stack_18.ipynb		price_quantizer_stack_18.ipynb
stack_01_1,4-5.ipynb		stack_01_1,4-5.ipynb
stack_02_1,3-5.ipynb		stack_02_1,3-5.ipynb
stack_03_1,3-7.ipynb		stack_03_1,3-7.ipynb
stack_04_1,3-7_model7-kfold.ipynb		stack_04_1,3-7_model7-kfold.ipynb
stack_04_1,3-7_model7-kfold_mean.ipynb		stack_04_1,3-7_model7-kfold_mean.ipynb
stack_05_1,3-7,9.ipynb		stack_05_1,3-7,9.ipynb
stack_06_1,3-7,9-10.ipynb		stack_06_1,3-7,9-10.ipynb
stack_08_1-7,9-11.ipynb		stack_08_1-7,9-11.ipynb
stack_09_1-11.ipynb		stack_09_1-11.ipynb
stack_10_1-11,14-15.ipynb		stack_10_1-11,14-15.ipynb
stack_11_1-11,14-17.ipynb		stack_11_1-11,14-17.ipynb
stack_12_1-18,20.ipynb		stack_12_1-18,20.ipynb
stack_lgb_14_1-22-1.ipynb		stack_lgb_14_1-22-1.ipynb
stack_mean_10_1-11,14-15.ipynb		stack_mean_10_1-11,14-15.ipynb
stack_mean_11_1-11,14-17.ipynb		stack_mean_11_1-11,14-17.ipynb
stack_mean_12_1-18,20.ipynb		stack_mean_12_1-18,20.ipynb
stack_mean_13_1-18,20-22.ipynb		stack_mean_13_1-18,20-22.ipynb
stack_mean_14_1-22.ipynb		stack_mean_14_1-22.ipynb
stack_mean_15_1-23,25-27.ipynb		stack_mean_15_1-23,25-27.ipynb
stack_mean_16_1-27.ipynb		stack_mean_16_1-27.ipynb
stack_mean_17_1-31.ipynb		stack_mean_17_1-31.ipynb
stack_mean_18_1-10,12-22,24-29.ipynb		stack_mean_18_1-10,12-22,24-29.ipynb
stack_mean_19_1-34,36-37.ipynb		stack_mean_19_1-34,36-37.ipynb
stack_mean_20_1-37.ipynb		stack_mean_20_1-37.ipynb
stack_mean_no_search_20_33,35-37.ipynb		stack_mean_no_search_20_33,35-37.ipynb
stack_mean_selected_ratios_20_1-37.ipynb		stack_mean_selected_ratios_20_1-37.ipynb
stack_parea_07_1,3-7,9-11.ipynb		stack_parea_07_1,3-7,9-11.ipynb

skyhuang1208/tbrain-realestate

Folders and files

Latest commit

History

Repository files navigation

TBrain - E.SUN House Price Prediction

Contributors

Achievement

Documentation

Structure

How it work

About

Resources

Stars

Watchers

Forks

Releases

Packages 0

Contributors 2

Languages

Packages