Brainscape's Knowledge GenomeTM

Browse over 1 million classes created by top students, professors, publishers, and experts.


See full index

Machine Learning > Lecture 7 - Features > Flashcards

Lecture 7 - Features Flashcards

1
Q

What are the 4 stages of data pre-processing

A
  1. Data cleaning
  2. Data integration
  3. Data reduction
  4. Data transformation
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
2
Q

What are features

A

features, also called attributes, are defined as mapping from the instance space to the feature domain.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
3
Q

What are the three main categories of feature statistics

A
  • Statistics of central tendency
  • Statistic of dispersion
  • Shape statistics
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
4
Q

What are the 3 main statistics of central tendency

A
  1. mean
  2. median
  3. mode
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
5
Q

What are the 2 statistics of dispersion

A
  1. Variance omega^2
  2. Standart deviation omega
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
6
Q

What are the statistics of dispersion

A
  • range
  • midrange point
  • quantiles
  • interguartile range
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
7
Q

The ____ is more sensetive to outlier than the ____

median or mean

A

mean
median

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
8
Q

what is skewness

A

Skewness is then defined as m/omega^3. A positive value of skewness means that the distribution is right-skewed, which means that the right tail is longer than the left tail. Negative skweness indicates the opposite.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
9
Q

What is Kurtosis

A

m/omega^4. People often use excess kurtosis m/omega^4 - 3. Positive excess kurtosis means that the distrubution is more sharply peaked than the normal distribution.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
10
Q

when can structured features be constructed

A
  1. prior to learning the model
  2. during learning the model
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
11
Q

What is normalisation

A

From Quantitave to Quantitative
Adapt the scale of quantitative features.

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
12
Q

What is calibration

A

From ortinal, categorical and boolean TO Quantitative
Adds a scale to features that don’t have one

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
13
Q

What is discretisation

A
  • from quantitative to ordinal
  • from quantitative to categorical
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
14
Q

what is ordering

A
  • from ordinal to ordinal
  • from categorical to ordinal
  • from boolean to ordinal
How well did you know this?
1
Not at all
2
3
4
5
Perfectly
15
Q

What is unordering

A

from ordinal to categorical

How well did you know this?
1
Not at all
2
3
4
5
Perfectly
16
Q

what is grouping

A

from categorical to categorical

17
Q

what is thresholding

A
  • from quantitative to boolean
  • from ordinal to boolean
18
Q

what is binarisation

A

from categorical to boolean

19
Q

Define thresholding. in words not table

A

Thresholding transforms a quantitave or an ordinal feature into a boolean feature by dinding a feature value to split on.

20
Q

how do we set the threshold for thresholding?

A
  • Supervised thresholding: hand picked for better performance
  • unsupervised thresholding: use centeral tendency statistics like mean/median
21
Q

Describe Discretisation

A

Discretisation transforms a quantitative feature into an ordinal feature, by creating bins where each bin is an interval

22
Q

name and exaplain 2 types of discretisation

A
  • supervised: bottom-up, work by progressively splitting bins
  • unsupervised: equal bin width, equal width discretisation
23
Q

Define normalisation

A

Feature normalisation neutralises the effect of different quantitative features being measured on different scales.

24
Q

Give to formulas with which we can normalise data

A
  • min-max
  • z-scores
25
Q

what is PCA

A

Principal component analysis is a feature-construbtion teqnique. It works by computing the principal components and using them to performs a change of basis on the data.

26
Q

Can PCA be performed on quantitative features?

A

yes

27
Q

What is the idea of pca

A

The idea of PCA is to find tehse correlcations and create a new feature that could be represented as a linear combination of the oringial features.

28
Q

in PCA, the sum of squared distances of projected points from the origin are called ____

A

eigenvalues

29
Q

What are principal components

A

principal components are new features constructed as a linear combination of original features

30
Q

give 2 approaches to extract principal components

A
  1. Singular value decomposition
  2. eigendecomposition
31
Q

How does singular value decomposition work

A

using matrixs rows for each feature.

32
Q

What is imputation

A

Imputation is the process of filling in missing data

33
Q

name 3 imputation techniques

A
  1. Mean imputation
  2. Regression imputation
  3. Expectation maximisation
34
Q

what is mean imputation

A

calculate the per class mean/median/mode

35
Q

what is regression imputation

A

a regression model is estimated to predict the observed vlaues of a variable based on other variables.

36
Q

what is expectation maximisation

A

assuming a multivariate model over all features, use the observed values for maximum-likelyhood estimation of the model parameters, then derive expectations for the unobserved feature values and interate.

Machine Learning (11 decks)

Brainscape helps you reach your goals faster, through stronger study habits.
© 2024 Bold Learning Solutions. Terms and Conditions