All Projects → alxmamaev → sdsj-automl

alxmamaev / sdsj-automl

Licence: other
Sberbank Data Science Jorney Auto-ML competition

Programming Languages

python
139335 projects - #7 most used programming language
shell
77523 projects

Projects that are alternatives of or similar to sdsj-automl

Auto viml
Automatically Build Multiple ML Models with a Single Line of Code. Created by Ram Seshadri. Collaborators Welcome. Permission Granted upon Request.
Stars: ✭ 216 (+671.43%)
Mutual labels:  automl
Ray
An open source framework that provides a simple, universal API for building distributed applications. Ray is packaged with RLlib, a scalable reinforcement learning library, and Tune, a scalable hyperparameter tuning library.
Stars: ✭ 18,547 (+66139.29%)
Mutual labels:  automl
oboe
An AutoML pipeline selection system to quickly select a promising pipeline for a new dataset.
Stars: ✭ 76 (+171.43%)
Mutual labels:  automl
Adatune
Gradient based Hyperparameter Tuning library in PyTorch
Stars: ✭ 226 (+707.14%)
Mutual labels:  automl
Tune Sklearn
A drop-in replacement for Scikit-Learn’s GridSearchCV / RandomizedSearchCV -- but with cutting edge hyperparameter tuning techniques.
Stars: ✭ 241 (+760.71%)
Mutual labels:  automl
mindsdb native
Machine Learning in one line of code
Stars: ✭ 34 (+21.43%)
Mutual labels:  automl
Pysr
Simple, fast, and parallelized symbolic regression in Python/Julia via regularized evolution and simulated annealing
Stars: ✭ 213 (+660.71%)
Mutual labels:  automl
Hypernets
A General Automated Machine Learning framework to simplify the development of End-to-end AutoML toolkits in specific domains.
Stars: ✭ 221 (+689.29%)
Mutual labels:  automl
Haq
[CVPR 2019, Oral] HAQ: Hardware-Aware Automated Quantization with Mixed Precision
Stars: ✭ 247 (+782.14%)
Mutual labels:  automl
NiaAML
Python automated machine learning framework.
Stars: ✭ 25 (-10.71%)
Mutual labels:  automl
Nascell Automl
Stars: ✭ 236 (+742.86%)
Mutual labels:  automl
Pocketflow
An Automatic Model Compression (AutoMC) framework for developing smaller and faster AI applications.
Stars: ✭ 2,672 (+9442.86%)
Mutual labels:  automl
managed ml systems and iot
Managed Machine Learning Systems and Internet of Things Live Lesson
Stars: ✭ 35 (+25%)
Mutual labels:  automl
Automlpipeline.jl
A package that makes it trivial to create and evaluate machine learning pipeline architectures.
Stars: ✭ 223 (+696.43%)
Mutual labels:  automl
S2E
Q. Yao, H. Yang, B. Han, G. Niu, J. Kwok. Searching to Exploit Memorization Effect in Learning from Noisy Labels. ICML 2020
Stars: ✭ 18 (-35.71%)
Mutual labels:  automl
Awesome Nas Papers
Awesome Neural Architecture Search Papers
Stars: ✭ 213 (+660.71%)
Mutual labels:  automl
AutoSpeech
[InterSpeech 2020] "AutoSpeech: Neural Architecture Search for Speaker Recognition" by Shaojin Ding*, Tianlong Chen*, Xinyu Gong, Weiwei Zha, Zhangyang Wang
Stars: ✭ 195 (+596.43%)
Mutual labels:  automl
AutoTS
Automated Time Series Forecasting
Stars: ✭ 665 (+2275%)
Mutual labels:  automl
FEDOT.Web
Graphic tool for the automated evolutionary design of composite models
Stars: ✭ 33 (+17.86%)
Mutual labels:  automl
nitroml
NitroML is a modular, portable, and scalable model-quality benchmarking framework for Machine Learning and Automated Machine Learning (AutoML) pipelines.
Stars: ✭ 40 (+42.86%)
Mutual labels:  automl

Sberbank AutoML solution

Dataset preparation

  • If the dataset is big (>2GB) then we calculate features correlation matrix and the delete correlated features
  • Else we make Mean Target Encoding and One Hot Encoding. 
  • After that, we select top-10 features by coefficients of the linear model (Ridge/LogisticRegression)
  • We generate new features by pair division from top-10 features. This method generates 90 new features (10^2–10) and concatenates it to the dataset.

Model training

  • If the dataset is small then we can train three LightGBM models by k-folds, after that blend prediction from every fold.
  • If the dataset is big and the time limit is small (5 minutes) then we just train linear models (logistic regression or ridge)
  • Else we train one big LightGBM (n_estimators=800)

Result

5th place on private leaderboard

Note that the project description data, including the texts, logos, images, and/or trademarks, for each open source project belongs to its rightful owner. If you wish to add or remove any projects, please contact us at [email protected].