• Verkeerd document? Gratis ruilen
  • Geschreven door studenten die geslaagd zijn
  • Direct beschikbaar na je betaling
  • Online lezen of als PDF
  • 4,6 TrustPilot
0
Verkopen
Kies je studieland
Kies je taal
Start met verkopen Account aanmaken
  • Universiteit Antwerpen (UA)
  • Handelsingenieur
  • Machine Learning for Business (2103TEWDAS)
Om documenten op je verlanglijstje te zetten, moet je ingelogd zijn
Document preview thumbnail
Voorbeeld 4 van de 93 pagina's
Samenvatting

Machine Learning for Business: samenvatting

0.0
(0 beoordelingen)
5 verkocht
93 pagina's
2025/2026 Geschreven voor schooljaar
Instelling
Universiteit Antwerpen (UA)
Vak
Machine Learning for Business (2103TEWDAS)
Document preview thumbnail
Voorbeeld 4 van de 93 pagina's

This summary is based on the slides covered in the Machine Learning for Business course taught at the Faculty of Business and Economics at the University of Antwerp. The theory covered in the tutorials is also discussed at the end of this summary.

Voorbeeld van de inhoud

MACHINE LEARNING FOR BUSINESS
2025-2026




This summary is based on the slides covered in the Machine Learning for Business course taught at the
Faculty of Business and Economics at the University of Antwerp.


1

,1: INTRO TO MACHINE LEARNING
WHAT IS…

1. Key concepts and terms
There are a lot of terms, but they don’t have fixed definitions. Each term can be used interchangeably depending on the
situation and environment. However, the following definitions will give you a guideline.


1.1 Data science
Data science = the process of extracting insights and knowledge from data using techniques from statistics, computer
science and domain expertise
●​ the broadest of terms
●​ used by people with backgrounds in statistics, computer science, mathematics, economics, engineering
or business analytics (~ everyone)


1.2 Data analytics
Data analytics = the process of examining, cleaning, transforming and modeling data to discover useful information and
support decision-making
●​ the most manual of the “data sciences”
●​ used by people with backgrounds in business, economics, marketing, finance or statistics


1.3 Data mining
Data mining = the process of discovering patterns, relationships or anomalies in large datasets using statistical and
computational methods
→ eg. patterns in products that people often buy together
●​ the oldest of the “data sciences”
●​ used when data analytics is not doable anymore, but you still make the decisions yourself
●​ used by people with backgrounds in business intelligence, marketing, statistic or computer science


1.4 Machine learning
Machine learning = the process of building algorithms that can learn from data and make predictions or decisions without
being explicitly programmed
→ eg. ML guesses a customer’s needs and decides if the customer needs a human employee to help with info
●​ the highest value of the “data sciences”
●​ decisions are made by the algorithm instead of a human
●​ used by people with backgrounds in computer science, data science, (digital business) engineering,
mathematics or physics (~ more technical and not mainly about business understanding)


1.5 Artificial intelligence
Artificial intelligence = the art of creating systems that can perform specific tasks typically requiring human intelligence,
such as reasoning, learning and perception
→ eg. AI for chatbot in customer service
●​ the hippest of the “data sciences”
●​ used by people with backgrounds in computer science, robotics, cognitive science or philosophy of
mind

2

,Artificial (General) Intelligence = the quest for a theoretical form of AI that can perform any intellectual task a human
can, with general reasoning and learning abilities across domains
●​ the absolute hippest of the “data sciences”
●​ not restricted to general capabilities or domain- or task specific knowledge
●​ most used by Sam Altman, Ilya Sutskever and other AI prophets


1.6 Big data
Big data = extremely large and complex datasets that require advanced tools and infrastructure to store, process and
analyze
●​ a bit of an outdated term in 2025 (we can store almost anything)
●​ used by people with backgrounds in computer science, information systems, cloud computing or data
engineering


1.7 Data engineering
Data engineering = a discipline of designing and building systems for collecting, storing and processing (big) data
efficiently and reliably
●​ underrated, but the number one reason why data science projects succeed or fail​
→ if you have a good model, the implementation is crucial for it to succeed
●​ used by people with backgrounds in software engineering, computer science or information technology


1.8 Machine learning engineering
Machine learning engineering = the practice of implementing, deploying and maintaining machine learning models in
production environments
●​ one of the hardest things to do well
●​ used by people with backgrounds in computer science, software engineering or applied mathematics

The world constantly changes, so tomorrow your model might not be as good at making predictions as it is today. This is a
problem we have only recently been confronted with. All models failed after COVID because this was a situation that we
had never seen before.
→ eg. prediction for stock amount in stores: physical stores (closed) had too much and online stores had too little stock


1.9 Data governance
Data governance = a set of policies, processes and standards that ensure data is accurate, secure and used responsibly
across an organization
→ eg. Is it ethical to use people’s ethnicity when building a fraud detection model?
●​ the most overlooked aspect of data science
●​ used by people with backgrounds in information systems, law, business administration or data
management

The more we use data on a day-to-day basis in our businesses, the more we need to take into account the risks, the things
that can go wrong and ways to safely store the data.


2. Time to exercise 🐾
pdf ‘1. Introduction to Machine Learning’
slides 34 - 40

3

, DATA TYPES

1. Key concepts

1.1 Structured data
Structured data = data that can be neatly displayed in rows and columns in relational databases
●​ easy to retrieve and analyze
●​ typically consists of numbers, dates, strings
●​ does not require much preprocessing before use in classifier

EXAMPLE: tabular data in a SQL database or spreadsheet
→ the students’ names are replaced by an identifier (pseudonymisation) to anonymize the data




1.2 Unstructured data
Unstructured data = data that is primarily meant for humans, not machines
●​ challenging to retrieve and analyze with machine learning models
●​ requires preprocessing before it can be used in a classifier
●​ makes up to +/- 80% of data out there

EXAMPLE: text, video, audio, images …


1.3 Semi-structured data
Semi-structured data = data that is somewhere in between structured and unstructured
→ has some structure, but structure is not rigid

EXAMPLE: HTML, XML, JSON, YAML …
●​ (L) missing data: some people might not fill in their city
●​ (L) other countries structure their addresses differently (districts, postal codes …)
●​ (R) all these things have a meaning, but you have to know the context to be able to understand them




4

Inhoudsopgave

  1. 01 MACHINE LEARNING FOR BUSINESS 1
  2. 02 2025-2026 1
  3. 03 1: INTRO TO MACHINE LEARNING 2
    1. WHAT IS… 2
  4. 04 1.1 Data science 2
  5. 05 1.2 Data analytics 2
  6. 06 1.3 Data mining 2
  7. 07 1.4 Machine learning 2
  8. 08 1.5 Artificial intelligence 2
  9. 09 1.6 Big data 3
  10. 10 1.7 Data engineering 3
  11. 11 1.8 Machine learning engineering 3
  12. 12 1.9 Data governance 3
    1. 2. Time to exercise ? 3
    2. DATA TYPES 4
  13. 13 1.1 Structured data 4
  14. 14 1.2 Unstructured data 4
  15. 15 1.3 Semi-structured data 4
    1. MODELLING 5
  16. 16 2.1 Features 5
  17. 17 2.2 Instances/samples 5
  18. 18 2.3 X 5
  19. 19 2.4 Labels/targets 6
  20. 20 2.5 y 6
  21. 21 2.6 Supervised learning 6
  22. 22 2.7 Unsupervised learning 7
  23. 23 2: THE CRISP-DM PROCESS 8
    1. THE CRISP-DM PROCESS 8
  24. 24 1.1 Business understanding 8
  25. 25 1.2 Data understanding 9
  26. 26 1.3 Data preparation 9
    1. DATASET PREPROCESSING 10
  27. 27 1.1 Sampling 10
  28. 28 1.2 Encoding 11
  29. 29 1.3 Missing values 13
  30. 30 1.4 Outliers 14
    1. THE CRISP-DM PROCESS 15
  31. 31 1.1 Modeling 15
  32. 32 1.2 Evaluation 15
  33. 33 1.3 Deployment 16
  34. 34 3: CLASSIFICATION & DECISION TREES 17
    1. CLASSIFICATION 17
  35. 35 1.1 Features 17
  36. 36 1.2 Instances 17
  37. 37 1.3 Labels/targets 17
  38. 38 1.4 Fitting a model 17
  39. 39 1.5 Supervised segmentation 18
  40. 40 1.6 Parametric learning 19
    1. DECISION TREES 20
  41. 41 1.1 Decision tree 20
  42. 42 1.2 Recursive partitioning 20
  43. 43 1.3 Splitting rule 21
  44. 44 1.4 Entropy 21
  45. 45 1.5 Information gain 21
  46. 46 1.6 GINI impurity 22
  47. 47 1.7 Stopping rule 23
  48. 48 1.8 Assignment rule 23
    1. GLOSSARY 23
  49. 49 4: LOGISTIC REGRESSION & OVERFITTING 24
    1. LOGISTIC REGRESSION 24
  50. 50 1.1 Parametric learning 24
  51. 51 1.2 Objective function / loss function 24
  52. 52 1.3 Gradient descent 24
  53. 53 1.4 Linear regression 24
  54. 54 1.5 Logistic function 25
  55. 55 1.6 Logistic regression 25
  56. 56 1.7 Binary cross-entropy 25
    1. OVERFITTING & GENERALIZATION 26
  57. 57 1.1 Training dataset 26
  58. 58 1.2 Generalization 26
  59. 59 1.3 Test dataset 27
  60. 60 1.4 Learning curve 27
  61. 61 1.5 Overfitting 28
  62. 62 1.6 Underfitting 29
  63. 63 1.7 Bias-variance tradeoff 30
  64. 64 1.8 Validation dataset 30
  65. 65 1.9 Single validation split 30
  66. 66 1.10 Cross-validation 31
  67. 67 1.11 Hyperparameter tuning 31
  68. 68 1.12 Early stopping 32
  69. 69 1.13 Regularization 32
    1. 2. Important: training, validation and test dataset 33
  70. 70 5: EVALUATING CLASSIFIERS 34
    1. EVALUATING CLASSIFICATION MODELS 34
  71. 71 2.1 Positives and negatives 34
  72. 72 2.2 Accuracy 34
  73. 73 2.3 Confusion matrix 34
  74. 74 2.4 Cost-benefit matrix 35
  75. 75 2.5 Expected profit 36
  76. 76 2.6 Baselines 36
    1. VISUALIZING MODEL PERFORMANCE 36
  77. 77 1.1 Ranking classifier 36
  78. 78 1.2 Profit curve 37
  79. 79 1.3 ROC curve 37
  80. 80 1.4 Area under curve (AUC) 38
  81. 81 1.5 Cumulative response curve 39
  82. 82 1.6 Lift curve 39
  83. 83 6: NAIVE BAYES & SVM 40
    1. NAIVE BAYES 40
  84. 84 1.1 Evidence 40
  85. 85 1.2 Joint probabilities and independence 40
  86. 86 1.3 Bayes rule 40
  87. 87 1.4 Naive Bayes and conditional independence 40
  88. 88 1.5 Evidence lift 41
    1. SUPPORT VECTOR MACHINES 43
  89. 89 1.1 Linearly separable dataset 43
  90. 90 1.2 Decision boundary 43
  91. 91 1.3 Support Vector Machine 43
  92. 92 1.4 Kernel trick/function 44
    1. RANDOM FORESTS 44
  93. 93 1.1 Ensembling 44
  94. 94 1.2 Random Forest 45
  95. 95 1.3 Bagging 45
  96. 96 1.4 Attribute sampling 45
  97. 97 7: UNSUPERVISED LEARNING - SIMILARITY, NEIGHBORS AND CLUSTERS 47
    1. SIMILARITIES 47
  98. 98 2.1 How do we measure similarity? 47
    1. NEIGHBORS 48
  99. 99 1.1 How to know which k to choose? 49
  100. 100 1.2 How to assign weights? 49
  101. 101 1.3 Uses of k-nearest neighbors 49
    1. CLUSTERS 50
  102. 102 8: TEXT MINING & RULE MINING 52
    1. TEXT MINING 52
  103. 103 2.1 Unstructured vs structured data 52
  104. 104 2.2 Text preprocessing 52
  105. 105 2.3 Terminology 53
  106. 106 2.4 Bag of words 53
  107. 107 2.5 TF-IDF 54
  108. 108 2.6 N-gram 55
  109. 109 2.7 Named entity recognition 55
  110. 110 2.8 Topic model 55
  111. 111 2.9 Word embedding 56
    1. ASSOCIATION RULE MINING 56
  112. 112 2.1 Transactions 56
  113. 113 2.2 Itemsets 57
  114. 114 2.3 Frequent itemsets 57
  115. 115 2.4 Association rules 57
  116. 116 2.5 Support 58
  117. 117 2.6 Confidence 58
  118. 118 2.7 Association Rule Mining: apriori algorithm 58
  119. 119 2.8 Lift 60
  120. 120 9: RECOMMENDER SYSTEMS 61
    1. RECOMMENDER SYSTEM 61
  121. 121 1.1 Recommender system 61
    1. 2. Baselines 61
  122. 122 2.1 Editorial (expert) recommendation 61
  123. 123 2.2 Popularity-based recommendation 62
  124. 124 2.3 Item-to-item recommendation 62
  125. 125 2.4 Social recommendation 63
    1. 3. Content-based and collaborative filtering 63
  126. 126 1.6 Personalized recommendation 63
  127. 127 1.7 Contextual recommendation 64
    1. RECOMMENDATION AS A ML PROBLEM 64
  128. 128 1.1 Behavioral data 64
  129. 129 1.2 Implicit feedback dataset 64
  130. 130 1.3 Dataset preprocessing 65
  131. 131 1.4 Training, validation, test dataset Training and hold-out test dataset 65
  132. 132 1.5 Predictions 66
  133. 133 1.6 Metrics 66
    1. RECOMMENDATION ALGORITHMS: DATA PERSPECTIVE 67
  134. 134 1.1 Baseline 67
  135. 135 1.2 Content-based recommendation 67
  136. 136 1.3 Collaborative filtering 68
  137. 137 1.4 UserKNN 69
  138. 138 1.5 ItemKNN 69
  139. 139 1.6 Hybrid recommendation 70
    1. UNDER THE HOOD 70
  140. 140 10: DEEP LEARNING 72
    1. NEURAL NETWORKS 72
  141. 141 1.1 The perceptron 72
  142. 142 1.2 Activation function 72
  143. 143 1.3 Multi-layer perceptron 73
  144. 144 1.4 Forward pass 75
  145. 145 1.5 Loss function 76
  146. 146 1.6 Backpropagation (backward pass) 76
  147. 147 1.7 Gradient Descent Algorithm 76
  148. 148 1.8 Stochastic Gradient Descent 77
    1. 2. Pros and cons of Neural Networks 77
    2. DEEP LEARNING 77
  149. 149 1.1 Convolutional Neural Networks (CNNs) 77
  150. 150 1.2 Recurrent Neural Networks (RNNs) 79
  151. 151 1.3 Autoencoders 80
  152. 152 1.4 Transformers 81
  153. 153 1.5 Foundation Models (Large Language Models) 82
  154. 154 11: ENSEMBLING 83
    1. ENSEMBLING 83
  155. 155 1.1 Ensembling 83
  156. 156 1.2 Bagging 83
  157. 157 1.3 Random Forest 83
  158. 158 1.4 Boosting 84
  159. 159 1.5 AdaBoost 84
  160. 160 1.6 Gradient Boosting 86
  161. 161 1.7 Stacking 87
  162. 162 EXTRA: MACHINE LEARNING AND DEEP LEARNING (tutorial 10) 88
    1. UNIVERSAL APPROXIMATION THEOREM 88
  163. 163 2.1 Example 88
  164. 164 2.2 Limitations 89
    1. DEEP LEARNING 89
  165. 165 1.1 Example 90
    1. 2. Why use Deep Learning? 91
  166. 166 2.1 Timeline 91
    1. 3. CNN - image classification with normal neural net 91
  167. 167 3.1 Classifying X’s and O’s 92
  168. 168 3.2 Convolutions: kernels 92
  169. 169 3.3 Creating a feature map 93
  170. 170 3.4 Pooling 93
  171. 171 3.5 CNN main takeaways 93

Documentinformatie

Studie
Handelsingenieur
Geüpload op
6 januari 2026
Aantal pagina's
93
Geschreven in
2025/2026
Type
Samenvatting
Meld schending auteursrecht
$16.13
Verkeerd document? Gratis ruilen Binnen 14 dagen na aankoop en voor het downloaden kan je een ander document kiezen. Je kan het bedrag gewoon opnieuw besteden.
Geschreven door studenten die geslaagd zijn
Direct beschikbaar na je betaling
Online lezen of als PDF
Seller avatar
De reputatie van een verkoper is gebaseerd op het aantal documenten dat iemand tegen betaling verkocht heeft en de beoordelingen die voor die items ontvangen zijn. Er zijn drie niveau’s te onderscheiden: brons, zilver en goud. Hoe beter de reputatie, hoe meer de kwaliteit van zijn of haar werk te vertrouwen is.
HIstudent2022
4.7
(31)
Verkocht
222
Volgers
4
Items
37
Laatst verkocht
3 uur geleden

Gerelateerde vakken

Onderwerpen

  • tew
  • handelsingenieur
  • samenvatting
  • machine learning for business
  • python for machine learning

Waarom studenten kiezen voor Stuvia

Gemaakt door medestudenten, geverifieerd door reviews

Kwaliteit die je kunt vertrouwen: geschreven door studenten die slaagden en beoordeeld door anderen die dit document gebruikten.

Niet tevreden? Kies een ander document

Geen zorgen! Je kunt voor hetzelfde geld direct een ander document kiezen dat beter past bij wat je zoekt.

Betaal zoals je wilt, start meteen met leren

Geen abonnement, geen verplichtingen. Betaal zoals je gewend bent via Bancontact, iDeal of creditcard en download je PDF-document meteen.

Student with book image

“Gekocht, gedownload en geslaagd. Zo eenvoudig kan het zijn.”

Alisha Student

Bezig met je bronvermelding?

Maak nauwkeurige citaten in APA, MLA en Harvard met onze gratis bronnengenerator.

Meer weten Bronvermelding maken
Bezig met je bronvermelding?

Veelgestelde vragen