Skip to main content
Ctrl+K
Logo image

Introduction

  • TD AutoML とは
    • TD AutoML のリソース
  • FAQ
  • Notebook 別スライド・ビデオ集

ml_datasets

  • ml_datasets とは
  • データセット解説
    • AutoGluon example dataset
    • Bank marketing dataset
    • in-vehicle coupon recommendation dataset
    • Online Retail dataset
    • Telco customer churn dataset
    • California house pricing dataset
    • Transition matrix from TD sample datasets
    • Monthly totals of international airline passengers
    • M4 Forecasting Competition Dataset
    • Sample dataset for Next Best Action notebook
    • DP6 random dataset for Multi-touch attribution modeling
    • Dermatology Database for Clustering
    • Credit Card Fraud Detection Dataset
    • Artificial Dataset for CLUTO, A Clustering Toolkit
    • Covtype dataset for multiclass classfication
    • 20 newsgroup dataset
    • Cosmetics online store dataset

AutoGluon

  • AutoGluon とは
  • Quick Start
    • 2値分類
    • 多値分類
    • 回帰
  • フルオプションリスト
    • gluon_train
    • gluon_predict
  • Audience Studio との連携
  • アウトプットテーブル
    • Leaderboard(gluon_train)
    • Feature Importance
  • 特徴量エンジニアリング
  • Notebook を読み解こう(gluon_train)
    • Loading data from TD
    • Explanatory Data Analysis (EDA)
    • Data Preprocessing
    • Training step
    • Model Fitting
    • Model inspection
  • モデルの評価指標
    • Distribution of Predicted Probabilities
    • Normarized Confusion Matrix(混同行列)
    • ROC 曲線と ROC-AUC
    • Precision-Recall 曲線 と PR_AUC
    • Distributions
    • Feature Importance
    • Shapley Values
  • Notebook を読み解こう(gluon_predict)
  • 失敗したモデルの例
  • より良いモデルを目指して
  • ML Experiment Tracking and Model Management
    • Track ML Experiments
    • Record Evaluation Results for each Model
    • config/params.yaml
    • ml_experiment.dig
    • ml_record_evaluation_results.dig
    • queries/track_experiment.sql
    • queries/get_latest_model.sql
    • queries/record_evaluation.sql
    • get_eval_score_from_leaderboard.sql
  • FAQ

Contents

previous

<no title>

Contents

By Treasure Academy

© Copyright 2020-2024 Treasure Data.