[Kaggle] 캐글 커리큘럼 (by. 이유한님)

kaggle.

Kaggle 스터디를 위한 커리큘럼입니다.
커리큘럼 참여에 있어 "처음부터 끝까지 3번씩 따라쓰고 이해하는 것"이 중요합니다.
그리고 이 과정을 통해 어떠한 Data인지, Project는 어떤 것인지, 어떤 학습이 되었는지
이해하는 것 또한 중요합니다.

이유한님께서 캐글 스터디 전용 커리큘럼을 정리해주셨다고 해서 사이트 또한 함께 공유합니다.

https://kaggle-kr.tistory.com/32

약간의 변경을 했으니, 커리큘럼을 필사적으로 필사하여 따라가봅시다. 

1. 표 데이터(Tabular data)

1-1. 표를 이용한 이진 분류(Binary classification)

1st level. Titanic: Machine Learning from Disaster

Titanic: Machine Learning from Disaster

• 1주차 : 타이타닉 튜토리얼 1 - Exploratory data analysis, visualization, machine learning
• 2주차 : EDA To Prediction(DieTanic)
• 3주차 : Titanic Top 4% with ensemble modeling
• 4주차 : Introduction to Ensembling/Stacking in Python

2nd level. Porto Seguro’s Safe Driver Prediction

Porto Seguro’s Safe Driver Prediction

• 5주차 : Data Preparation & Exploration
• 6주차 : Interactive Porto Insights - A Plot.ly Tutorial
• 7주차 : XGBoost CV (LB .284)
• 8주차 : Porto Seguro Exploratory Analysis and Prediction

3rd level. Home Credit Default Risk

Home Credit Default Risk

• 9주차 : Introduction: Home Credit Default Risk Competition
• 10주차 : Introduction to Manual Feature Engineering
• 11주차 : Stacking Test-Sklearn, XGBoost, CatBoost, LightGBM
• 12주차 : LightGBM 7th place solution

1-2. 표를 이용한 다중 분류(Multi-class classification)

1st level. Costa Rican Household Poverty Level Prediction

Costa Rican Household Poverty Level Prediction

• 13주차 : A Complete Introduction and Walkthrough
• 14주차 : 3250feats->532 feats using shap[LB: 0.436]
• 15주차 : XGBoost

1-3. 표를 이용한 회귀(Regression)

1st level. New York City Taxi Trip Duration

New York City Taxi Trip Duration

• 16주차 : Dynamics of New York city - Animation
• 17주차 : EDA + Baseline Model
• 18주차 : Beat the benchmark!

2nd level. Zillow Prize: Zillow’s Home Value Prediction (Zestimate)

Zillow Prize: Zillow’s Home Value Prediction (Zestimate)

• 19주차 : Simple Exploration Notebook - Zillow Prize
• 20주차 : Simple XGBoost Starter (~0.0655)
• 21주차 : Zillow EDA On Missing Values & Multicollinearity
• 22주차 : XGBoost, LightGBM, and OLS and NN

2. 이미지 데이터(Image classification)

2-1. 이미지를 이용한 이진 분류(Binary classification)

1st level. Statoil/C-CORE Iceberg Classifier Challenge

Statoil/C-CORE Iceberg Classifier Challenge

• 23주차 : Keras Model for Beginners (0.210 on LB)+EDA+R&D
• 24주차 : Transfer Learning with VGG-16 CNN+AUG LB 0.1712
• 25주차 : Submarineering.EVEN BETTER PUBLIC SCORE until now.
• 26주차 : Keras+TF LB 0.18

2-2. 이미지를 이용한 다중 분류(Multi-class classification)

1st level. TensorFlow Speech Recognition Challenge

TensorFlow Speech Recognition Challenge

• 27주차 : Speech representation and data exploration
• 28주차 : Light-Weight CNN LB 0.74
• 29주차 : WavCeption V1: a 1-D Inception approach (LB 0.76)

3. 자연어 처리(Natural language processing)

1st level. Spooky Author Identification

Spooky Author Identification

• 30주차 : Spooky NLP and Topic Modelling tutorial
• 31주차 : Approaching (Almost) Any NLP Problem on Kaggle
• 32주차 : Simple Feature Engg Notebook - Spooky Author

2nd level. Mercari Price Suggestion Challenge

Mercari Price Suggestion Challenge

• 33주차 : Mercari Interactive EDA + Topic Modelling
• 34주차 : A simple nn solution with Keras (~0.48611 PL)
• 35주차 : Ridge (LB 0.41943)
• 36주차 : LGB and FM [18th Place - 0.40604]

3rd level. Toxic Comment Classification Challenge

Toxic Comment Classification Challenge

• 37주차 : [For Beginners] Tackling Toxic Using Keras
• 38주차 : Stop the S@#$ - Toxic Comments EDA
• 39주차 : Logistic regression with words and char n-grams
• 40주차 : Classifying multi-label comments (0.9741 lb)

4. 딥러닝을 이용한 객체 분할(Object segmentation)

1st level. 2018 Data Science Bowl

2018 Data Science Bowl

• 41주차 : Teaching notebook for total imaging newbies
• 42주차 : Keras U-Net starter - LB 0.277
• 43주차 : Nuclei Overview to Submission

5. 기타 : 이상 검출(anomaly detection), 시각화(visualization)

1st level. Credit Card Fraud Detection

Credit Card Fraud Detection

• 44주차 : In depth skewed data classif. (93% recall acc now)
• 45주차 : Anomaly Detection - Credit Card Fraud Analysis
• 46주차 : Semi-Supervised Anomaly Detection Survey

2nd level. Kaggle Machine Learning & Data Science Survey 2017

2017 Kaggle ML & DS Survey

• 47주차 : Novice to Grandmaster
• 48주차 : What do Kagglers say about Data Science ?
• 49주차 : PLOTLY TUTORIAL - 1