Uitgebreide samenvatting per lecture + Key concepts (kleinere
samenvatting van samenvatting)
Data Science lecture 1 5
Research Paradigms 5
Data Challenges 5
Application domain 5
Task definition questions 6
Supervised vs Unsupervised 6
Addressing data science problems: 7
Mean vs Median 7
Outliers 7
Regression 8
Simple linear regression 8
Multiple linear regression 8
Logistic Regression 9
Loss functions 9
Sigmoid 10
Lecture 2 11
Visualisation 11
Anscombe’s quartet 11
Visualisation Metaphors 11
Bad visualisation 12
Lecture 3 13
Supervised learning 13
Classification 13
Classification models 14
Vector space model 14
K-Nearest Neighbour (KNN) 14
Support vector machine(SVM) 15
Neural networks 16
Hidden layers 16
ReLU VS sigmoid 17
Single neurons 17
Cost functions 18
Gradient descent 18
Perceptrons 19
XOR problem 20
Feed forward networks 20
Training Neural nets 21
Lecture 4 22
Experiment setup 22
1
Jesse de Gans
, Hyper parameter tuning 22
Regression evaluation 22
Evaluation of rankings 22
Evaluation for classification 23
F-score 23
Classifier quality & analysis 23
Lecture 5 24
Network science 24
Network types 24
Real-world network properties 24
Network density 25
Degree 25
Components 25
Distance 25
Clustering coefficients 25
Centrality 26
Degree centrality 26
Closeness centrality 26
Betweenness centrality 26
Communities 27
Modularity maximisation 27
Lecture 6 28
Data collection 28
Using Existing labelled data 28
Create new labelled data 28
Inter-rater agreement 29
Interpretation of Cohen’s Kappa 29
Lecture 7 30
Data Preparation 30
Feature extraction 30
Dense vs Sparse data 30
Text Classification 31
Traditionally 31
Preprocessing: Raw text to features 32
Clean up and normalisation 32
Tokenization 32
Pre-processing with NLP tools 32
Feature creation 32
Image to matrix 33
Image feature extraction 33
Convolutional neural networks 33
Need to knows 34
Image preprocessing 34
2
Jesse de Gans
,Lecture 8 35
Choosing models and methods 35
Choosing supervised vs Unsupervised: 35
Choosing between classification clustering or regression: 35
Decide on features 35
Choosing the right estimator 35
Supervised Classification models 36
Transfer learning 36
Transfer learning for images 36
Transfer learning for text 36
Lecture 9 37
Feature normalisation 37
Scaling numerical features 37
Dimensionality reduction 37
PCA (Principal component Analysis) 38
Significance testing 38
Which test to use 38
Lecture 10 39
Natural Language processing 39
Text data challenges 39
Zipfs law 39
Bag-of-words model: Text as classification object 40
Words(terms) as features 40
Computing term weights (real valued) 40
Term frequency (tf) 40
Inverse document frequency (idf) 41
Tf-idf(term-frequency Inverse document frequency) 41
Term-document matrix 41
Words and polysemy 42
Word embeddings 42
Learning word embeddings 42
Neural language models 43
Application of transfer learning to image and text data 43
Lecture 11 44
Evaluation of classification 44
Evaluation for regression 44
Confusion matrices 44
Error analyses 45
Dimensionality reduction 46
Class imbalance 46
Machine learning 46
Hyper param optimization 47
3
Jesse de Gans
, Overfitting 47
Cross validation 47
Leave-one-out cross validation 48
Lecture 12 49
Big data 49
Responsible data science 49
Risks and opportunities 49
Explainable models 50
Key concepts: 51-61
4
Jesse de Gans
samenvatting van samenvatting)
Data Science lecture 1 5
Research Paradigms 5
Data Challenges 5
Application domain 5
Task definition questions 6
Supervised vs Unsupervised 6
Addressing data science problems: 7
Mean vs Median 7
Outliers 7
Regression 8
Simple linear regression 8
Multiple linear regression 8
Logistic Regression 9
Loss functions 9
Sigmoid 10
Lecture 2 11
Visualisation 11
Anscombe’s quartet 11
Visualisation Metaphors 11
Bad visualisation 12
Lecture 3 13
Supervised learning 13
Classification 13
Classification models 14
Vector space model 14
K-Nearest Neighbour (KNN) 14
Support vector machine(SVM) 15
Neural networks 16
Hidden layers 16
ReLU VS sigmoid 17
Single neurons 17
Cost functions 18
Gradient descent 18
Perceptrons 19
XOR problem 20
Feed forward networks 20
Training Neural nets 21
Lecture 4 22
Experiment setup 22
1
Jesse de Gans
, Hyper parameter tuning 22
Regression evaluation 22
Evaluation of rankings 22
Evaluation for classification 23
F-score 23
Classifier quality & analysis 23
Lecture 5 24
Network science 24
Network types 24
Real-world network properties 24
Network density 25
Degree 25
Components 25
Distance 25
Clustering coefficients 25
Centrality 26
Degree centrality 26
Closeness centrality 26
Betweenness centrality 26
Communities 27
Modularity maximisation 27
Lecture 6 28
Data collection 28
Using Existing labelled data 28
Create new labelled data 28
Inter-rater agreement 29
Interpretation of Cohen’s Kappa 29
Lecture 7 30
Data Preparation 30
Feature extraction 30
Dense vs Sparse data 30
Text Classification 31
Traditionally 31
Preprocessing: Raw text to features 32
Clean up and normalisation 32
Tokenization 32
Pre-processing with NLP tools 32
Feature creation 32
Image to matrix 33
Image feature extraction 33
Convolutional neural networks 33
Need to knows 34
Image preprocessing 34
2
Jesse de Gans
,Lecture 8 35
Choosing models and methods 35
Choosing supervised vs Unsupervised: 35
Choosing between classification clustering or regression: 35
Decide on features 35
Choosing the right estimator 35
Supervised Classification models 36
Transfer learning 36
Transfer learning for images 36
Transfer learning for text 36
Lecture 9 37
Feature normalisation 37
Scaling numerical features 37
Dimensionality reduction 37
PCA (Principal component Analysis) 38
Significance testing 38
Which test to use 38
Lecture 10 39
Natural Language processing 39
Text data challenges 39
Zipfs law 39
Bag-of-words model: Text as classification object 40
Words(terms) as features 40
Computing term weights (real valued) 40
Term frequency (tf) 40
Inverse document frequency (idf) 41
Tf-idf(term-frequency Inverse document frequency) 41
Term-document matrix 41
Words and polysemy 42
Word embeddings 42
Learning word embeddings 42
Neural language models 43
Application of transfer learning to image and text data 43
Lecture 11 44
Evaluation of classification 44
Evaluation for regression 44
Confusion matrices 44
Error analyses 45
Dimensionality reduction 46
Class imbalance 46
Machine learning 46
Hyper param optimization 47
3
Jesse de Gans
, Overfitting 47
Cross validation 47
Leave-one-out cross validation 48
Lecture 12 49
Big data 49
Responsible data science 49
Risks and opportunities 49
Explainable models 50
Key concepts: 51-61
4
Jesse de Gans
