Interpretable Machine Learning Applications: Part 1

提供:
Coursera Project Network
このガイド付きプロジェクトでは、次のことを行います。

How to select and compare different prediction models (classification regressors) for a real world dataset (FIFA 2018 Soccer World Cup Statistics).

How to extract the most important features, which impact the classifiers, in a model-agnostic approach, together with caveats.

How to get an insight into the way values of the most important features impact the predictions made by the classifiers.

Clock2-hour course, including time of video recordings, practicing and readings, taking the quiz.
Beginner初級
Cloudダウンロード不要
Video分割画面ビデオ
Comment Dots英語
Laptopデスクトップのみ

In this 1-hour long project-based course, you will learn how to create interpretable machine learning applications on the example of two classification regression models, decision tree and random forestc classifiers. You will also learn how to explain such prediction models by extracting the most important features and their values, which mostly impact these prediction models. In this sense, the project will boost your career as Machine Learning (ML) developer and modeler in that you will be able to get a deeper insight into the behaviour of your ML model. The project will also benefit your career as a decision maker in an executive position, or consultant, interested in deploying trusted and accountable ML applications. Note: This course works best for learners who are based in the North America region. We’re currently working on providing the same experience in other regions.

あなたが開発するスキル

  • Python basic knowledge
  • Features engineering
  • Machine learning classification (regression) models

ステップバイステップで学習します

ワークエリアを使用した分割画面で再生するビデオでは、講師がこれらの手順を説明します。

  1. Setting the stage (Python Jupyter Lab web-based Server environment, importing the dataset and file to train and test the designated classification regressors as prediction models).

  2. Train, test and estimate the accuracy (confusion matrix) of a Decision Tree classifier.

  3. Train, test and estimate the accuracy (confusion matrix) of a Random Tree classifier as an alternative to the previous one.

  4. Extract a ranking list of the features, which are most important for each one of our prediction models.

  5. Extract and plot the impact of the values of selected important features on predictions being made by each one of our prediction models.

ガイド付きプロジェクトの仕組み

ワークスペースは、ブラウザに完全にロードされたクラウドデスクトップですので、ダウンロードは不要です

分割画面のビデオで、講師が手順ごとにガイドします

よくある質問

よくある質問

さらに質問がある場合は、受講者ヘルプセンターにアクセスしてください。