Introduction

Figure 1

Poorly fitted data

Data preparation

Figure 1

Train and test set

Learning

Figure 1

Model training

Figure 2

Distance from target

Figure 3

Mean squared error

Figure 4

Model training

Modelling

Figure 1

Linear regression with binary outcome

Figure 2

Sigmoid function

Figure 3

  • You can reuse the Linear Regression code above and simply change LinearRegression() to LogisticRegression(). You should see a plot similar to the one below: Logistic regression

    • Figure 4

  • You can reuse the Linear Regression code above and simply change the model again to DecisionTreeClassifier(). You should see a plot similar to the one below: Logistic regression

    • Validation

    Figure 1

    Ren Hayakawa Archery Olympics

    Figure 2

    Validation set

    Figure 3

    5-fold validation

    Evaluation

    Figure 1

    Confusion matrix for logistic regressionConfusion matrix for decision trees

    Figure 2

    AUROCAUROC

    Bootstrapping

    Figure 1

    Bootstrapped accuracy

    Figure 2

    Bootstrapped accuracy with confidence

    Data leakage

    Figure 1

    Dataset leakage