 MLDA an der RWTH Aachen | Karteikarten & Zusammenfassungen

# Lernmaterialien für MLDA an der RWTH Aachen

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen MLDA Kurs an der RWTH Aachen zu.

TESTE DEIN WISSEN

What is the objective of Classifying?

Lösung anzeigen
TESTE DEIN WISSEN

- Establishing a decision surface seperating two or more classes

- the decision surface generalizes how to classify new data points

Lösung ausblenden
TESTE DEIN WISSEN

How do you estimate or choose the Regression coefficients (betha1 and betha2)?

Lösung anzeigen
TESTE DEIN WISSEN

To maximize the Maximum likelihood method

Lösung ausblenden
TESTE DEIN WISSEN

What does Discriminant analysis uses the training observations for?

Lösung anzeigen
TESTE DEIN WISSEN

- to determine the location of a boundary between the response classes

- observations from each class are treated as samples of a multidimensional normal distribution

Lösung ausblenden
TESTE DEIN WISSEN

Are the classification algorithms LR, LDA, QDA, KNN and SVM supervised or unsupervised methods?

Lösung anzeigen
TESTE DEIN WISSEN

supervised

Lösung ausblenden
TESTE DEIN WISSEN

Characteristics of response variables of classification problems

Lösung anzeigen
TESTE DEIN WISSEN

Response variable is qualitative / categorial

Lösung ausblenden
TESTE DEIN WISSEN

Why "naives" for the Naives Bayes?

Lösung anzeigen
TESTE DEIN WISSEN

Assumption: conditional independence for every feature

Lösung ausblenden
TESTE DEIN WISSEN

What is the drawback of the Naive Bayes?

Lösung anzeigen
TESTE DEIN WISSEN

The prediction might be poor, if some features depend on each other

Lösung ausblenden
TESTE DEIN WISSEN

Why is the linear regression not  suitable for classification problems? (2)

Lösung anzeigen
TESTE DEIN WISSEN

1. Problematic when the response variable doesn't take a natural ordering

2. LR proceeds response values bigger than 1 --> problematic to interpret as probability

Lösung ausblenden
TESTE DEIN WISSEN

How to solve confounding?

Lösung anzeigen
TESTE DEIN WISSEN

Multiple logistic regression

Lösung ausblenden
TESTE DEIN WISSEN

What is the strength of LR?

Lösung anzeigen
TESTE DEIN WISSEN

Typically works well even for correlated variables

Lösung ausblenden
TESTE DEIN WISSEN

Characterize the decision boundary for high K.

Low or high Variance and Bias?

Lösung anzeigen
TESTE DEIN WISSEN

- decision boundary is less flexible (close to linear)

- low variance

- high bias

Lösung ausblenden
TESTE DEIN WISSEN

When to use linear discriminant analysis (LDA) or quadratic discriminant analysis (QDA)?

Lösung anzeigen
TESTE DEIN WISSEN

- LDA: Covariance matrix assumed equal for all classes (linear boundaries)

Lösung ausblenden • 197536 Karteikarten
• 4646 Studierende
• 310 Lernmaterialien

## Beispielhafte Karteikarten für deinen MLDA Kurs an der RWTH Aachen - von Kommilitonen auf StudySmarter erstellt!

Q:

What is the objective of Classifying?

A:

- Establishing a decision surface seperating two or more classes

- the decision surface generalizes how to classify new data points

Q:

How do you estimate or choose the Regression coefficients (betha1 and betha2)?

A:

To maximize the Maximum likelihood method

Q:

What does Discriminant analysis uses the training observations for?

A:

- to determine the location of a boundary between the response classes

- observations from each class are treated as samples of a multidimensional normal distribution

Q:

Are the classification algorithms LR, LDA, QDA, KNN and SVM supervised or unsupervised methods?

A:

supervised

Q:

Characteristics of response variables of classification problems

A:

Response variable is qualitative / categorial

Q:

Why "naives" for the Naives Bayes?

A:

Assumption: conditional independence for every feature

Q:

What is the drawback of the Naive Bayes?

A:

The prediction might be poor, if some features depend on each other

Q:

Why is the linear regression not  suitable for classification problems? (2)

A:

1. Problematic when the response variable doesn't take a natural ordering

2. LR proceeds response values bigger than 1 --> problematic to interpret as probability

Q:

How to solve confounding?

A:

Multiple logistic regression

Q:

What is the strength of LR?

A:

Typically works well even for correlated variables

Q:

Characterize the decision boundary for high K.

Low or high Variance and Bias?

A:

- decision boundary is less flexible (close to linear)

- low variance

- high bias

Q:

When to use linear discriminant analysis (LDA) or quadratic discriminant analysis (QDA)?

A:

- LDA: Covariance matrix assumed equal for all classes (linear boundaries)   