Business Intelligence an der Universität Erlangen-Nürnberg | Karteikarten & Zusammenfassungen

Lernmaterialien für Business Intelligence an der Universität Erlangen-Nürnberg

Greife auf kostenlose Karteikarten, Zusammenfassungen, Übungsaufgaben und Altklausuren für deinen Business Intelligence Kurs an der Universität Erlangen-Nürnberg zu.

TESTE DEIN WISSEN

What are the different measure of the descriptive statistics ?

Lösung anzeigen
TESTE DEIN WISSEN

Measure 

Definition

Application

Mean

Weighted center (influences by values of every entry)

Quantitative values (1st choice)

Median

Weighted center (middle number of ordered values)

Quantitative values & extreme/ many outliers

Mode

Most frequently occurring value

Categorical values

Variance

Indicator for dispersion around the mean

 

Standard deviation

Square root of the variance

 

Correlation (coefficient)

Linear relationship between two numeriacal values (-1 - +1)

Data preparation – dimensionality reduction – eliminate highly correlated attributes

Lösung ausblenden
TESTE DEIN WISSEN

What are main data-preporcessing tasks ? Please briefly name the core tasks of each step ? 

Lösung anzeigen
TESTE DEIN WISSEN
  1. Data Cleaning
    • Missing values 
    • noisy data 
  2. Data Integration
    • integrate various sources/tables in one view/ table by using a query
  3. Data Reduction 
    • attribute subset selection 
    • Dimensionality reduction 
  4. Data Transformation
    • Aggregation 
    • Normalization 
    • Discretization
    • Feature construction/ Feature engineering 
Lösung ausblenden
TESTE DEIN WISSEN

Whats an outlier ? Why is it important to look do a proper oulier handling ? 

What techniques can you use for Detection

Lösung anzeigen
TESTE DEIN WISSEN

Outlier

Observation that does not fit to the others, numerically distant from the other points

Reasons

- Data Qualtity problems (e.g. typos/ wrong measurements)

- Exceptional or unusual situations (rare behaviour)

Handling

Exclusion, not representative for most of the data points, misleading values for the mean, ensure that the model can generalize (not to specific)

Detection Techniques

- Box Plots (quartile range based)
- DBSCAN (Clustering-based method)

Lösung ausblenden
TESTE DEIN WISSEN

Please give a brief defintion of the term "Artificial Intelligence" ?

Lösung anzeigen
TESTE DEIN WISSEN

- study/ science of making a computer (program)  performing tasks by using some kind of humanlike intelligence 

-making computer performing things at which people are better

Lösung ausblenden
TESTE DEIN WISSEN

What are stages of AI ? What stage are we currently in ? 

Lösung anzeigen
TESTE DEIN WISSEN
  • ANI, Artificial Narrow Intelligence (weak)
    •  machines perform specific tasks better than a human,
    • simple tasks e.g. chat bot
    • currently in
  • - Artificial General Intelligence (strong) 
    • think perform like a human
    • not reached yet (decades)
  • Artificial Super Intelligence
    • Surpass the human beeing
    • Far Of speculative

Lösung ausblenden
TESTE DEIN WISSEN

Please do a quick classification of intelligence taks ? 

Lösung anzeigen
TESTE DEIN WISSEN
  • basic tasks
    • perception, common sense, natural language processing, reasoning
  • intermediate tasks
    • Mathematics, games
  • expert tasks
    • financial analysis, engineering,


!! basic tasks are in terms of AI the more advanced tasks !!

Lösung ausblenden
TESTE DEIN WISSEN

What are the fields of Application in AI ? (AI Branches)

Lösung anzeigen
TESTE DEIN WISSEN
  • Robtics 
  • Speech (speech to text, text to speech)
  • vision (image recognition, machine vision) 
  • Machine Learning
  • expert systems (providing answers)
  • Natural Language Processing (classification, machine translation, question answering, text generation)
Lösung ausblenden
TESTE DEIN WISSEN

Please give a short defintion/ explanation of the "Machine Learning" term! 

Lösung anzeigen
TESTE DEIN WISSEN
  • Study of enabling computers to learn without being explicitly programmed 
  • Using algorithms to learn the intelligence giving them examples
  • Gradually learning & improve the accuracy (imitating the humans learning process)
  • Subdiscipline of Artificial Intelligence
Lösung ausblenden
TESTE DEIN WISSEN

What are stages in the Machine Learning process ?

Lösung anzeigen
TESTE DEIN WISSEN

1. Data preparation

2. Model building 

3. Model evaluation

4. Model optimizaition

5. Prediction (Model deployment) 


!! iterative nature - e.g.  data shift etc. !!

Lösung ausblenden
TESTE DEIN WISSEN

What are the different types of Machine Learning ?

Lösung anzeigen
TESTE DEIN WISSEN
  • Supervised 
    • Labelled data, goal: prediction task (reg/ class)
  • Unsupervised
    • Unlabeled data, goal: determine patterns groups
    • e.g. clustering, association
  • Semi-supervised
    • Using labelled & unlabelled training data
    • Few labelled instances to train and then label the unlabelled data using initial model then adding to labelled data begin process again (more and more data) 
  • Reinforcement
    • Using a system where an agent is getting a feedback (reward) for performing actions in a defined actions scope in a specific/ defined environment 
    • System learns with the trail&error method by maximizing the reward  
Lösung ausblenden
TESTE DEIN WISSEN

Briefly explain the goal of the first step of the first phase of the CRISP-DM process. 

What can be the problems which can occour by not executing this phase properly? What can be done to prevent the problems ? 

Lösung anzeigen
TESTE DEIN WISSEN

Business Understanding

Goal

1. Cleary enunciate the project objectives in terms of the business

2. translate into a proper data mining definition

3. prepare preliminary strategy


Problems/ problem sources (Project Owner perspective vs. Data Analyst perspective) 

  • - Communication 
    • technical terms (for PO) vs business terms for (DA) 
  • Lack of Understanding 
    • PO doesn't know the capabilties of the DA/ models (other expectations initally) vs DA doesn't undersstand how to help PO
  • Organization
    • adoption of requirements because of problems with data in later stages vs. PO is not engaged enough in the project


Solution/ Optimization Approaches

  • Bridging gap between DA/ PO in terms of Business Understanding using tools like cognitive maps 
  • define primary objectives (critical thinking, question them, precise enough? feasibility
  • determine analytics goals 
    • interpretability (no black box, management) 
    • reproducibility/ stability of the analysis 
    •  runtime (restriction time, ressources) 
    • interestingeness of outcome ("surprise" domain expert)
Lösung ausblenden
TESTE DEIN WISSEN

What is noisy data ? What tools can you use to detect it? 

Lösung anzeigen
TESTE DEIN WISSEN

--> random error in a variable. Results in a specific variance 


tools: 

Boxplot  - Binning - Regression - Clustering 


Lösung ausblenden
  • 154240 Karteikarten
  • 2902 Studierende
  • 117 Lernmaterialien

Beispielhafte Karteikarten für deinen Business Intelligence Kurs an der Universität Erlangen-Nürnberg - von Kommilitonen auf StudySmarter erstellt!

Q:

What are the different measure of the descriptive statistics ?

A:

Measure 

Definition

Application

Mean

Weighted center (influences by values of every entry)

Quantitative values (1st choice)

Median

Weighted center (middle number of ordered values)

Quantitative values & extreme/ many outliers

Mode

Most frequently occurring value

Categorical values

Variance

Indicator for dispersion around the mean

 

Standard deviation

Square root of the variance

 

Correlation (coefficient)

Linear relationship between two numeriacal values (-1 - +1)

Data preparation – dimensionality reduction – eliminate highly correlated attributes

Q:

What are main data-preporcessing tasks ? Please briefly name the core tasks of each step ? 

A:
  1. Data Cleaning
    • Missing values 
    • noisy data 
  2. Data Integration
    • integrate various sources/tables in one view/ table by using a query
  3. Data Reduction 
    • attribute subset selection 
    • Dimensionality reduction 
  4. Data Transformation
    • Aggregation 
    • Normalization 
    • Discretization
    • Feature construction/ Feature engineering 
Q:

Whats an outlier ? Why is it important to look do a proper oulier handling ? 

What techniques can you use for Detection

A:

Outlier

Observation that does not fit to the others, numerically distant from the other points

Reasons

- Data Qualtity problems (e.g. typos/ wrong measurements)

- Exceptional or unusual situations (rare behaviour)

Handling

Exclusion, not representative for most of the data points, misleading values for the mean, ensure that the model can generalize (not to specific)

Detection Techniques

- Box Plots (quartile range based)
- DBSCAN (Clustering-based method)

Q:

Please give a brief defintion of the term "Artificial Intelligence" ?

A:

- study/ science of making a computer (program)  performing tasks by using some kind of humanlike intelligence 

-making computer performing things at which people are better

Q:

What are stages of AI ? What stage are we currently in ? 

A:
  • ANI, Artificial Narrow Intelligence (weak)
    •  machines perform specific tasks better than a human,
    • simple tasks e.g. chat bot
    • currently in
  • - Artificial General Intelligence (strong) 
    • think perform like a human
    • not reached yet (decades)
  • Artificial Super Intelligence
    • Surpass the human beeing
    • Far Of speculative

Mehr Karteikarten anzeigen
Q:

Please do a quick classification of intelligence taks ? 

A:
  • basic tasks
    • perception, common sense, natural language processing, reasoning
  • intermediate tasks
    • Mathematics, games
  • expert tasks
    • financial analysis, engineering,


!! basic tasks are in terms of AI the more advanced tasks !!

Q:

What are the fields of Application in AI ? (AI Branches)

A:
  • Robtics 
  • Speech (speech to text, text to speech)
  • vision (image recognition, machine vision) 
  • Machine Learning
  • expert systems (providing answers)
  • Natural Language Processing (classification, machine translation, question answering, text generation)
Q:

Please give a short defintion/ explanation of the "Machine Learning" term! 

A:
  • Study of enabling computers to learn without being explicitly programmed 
  • Using algorithms to learn the intelligence giving them examples
  • Gradually learning & improve the accuracy (imitating the humans learning process)
  • Subdiscipline of Artificial Intelligence
Q:

What are stages in the Machine Learning process ?

A:

1. Data preparation

2. Model building 

3. Model evaluation

4. Model optimizaition

5. Prediction (Model deployment) 


!! iterative nature - e.g.  data shift etc. !!

Q:

What are the different types of Machine Learning ?

A:
  • Supervised 
    • Labelled data, goal: prediction task (reg/ class)
  • Unsupervised
    • Unlabeled data, goal: determine patterns groups
    • e.g. clustering, association
  • Semi-supervised
    • Using labelled & unlabelled training data
    • Few labelled instances to train and then label the unlabelled data using initial model then adding to labelled data begin process again (more and more data) 
  • Reinforcement
    • Using a system where an agent is getting a feedback (reward) for performing actions in a defined actions scope in a specific/ defined environment 
    • System learns with the trail&error method by maximizing the reward  
Q:

Briefly explain the goal of the first step of the first phase of the CRISP-DM process. 

What can be the problems which can occour by not executing this phase properly? What can be done to prevent the problems ? 

A:

Business Understanding

Goal

1. Cleary enunciate the project objectives in terms of the business

2. translate into a proper data mining definition

3. prepare preliminary strategy


Problems/ problem sources (Project Owner perspective vs. Data Analyst perspective) 

  • - Communication 
    • technical terms (for PO) vs business terms for (DA) 
  • Lack of Understanding 
    • PO doesn't know the capabilties of the DA/ models (other expectations initally) vs DA doesn't undersstand how to help PO
  • Organization
    • adoption of requirements because of problems with data in later stages vs. PO is not engaged enough in the project


Solution/ Optimization Approaches

  • Bridging gap between DA/ PO in terms of Business Understanding using tools like cognitive maps 
  • define primary objectives (critical thinking, question them, precise enough? feasibility
  • determine analytics goals 
    • interpretability (no black box, management) 
    • reproducibility/ stability of the analysis 
    •  runtime (restriction time, ressources) 
    • interestingeness of outcome ("surprise" domain expert)
Q:

What is noisy data ? What tools can you use to detect it? 

A:

--> random error in a variable. Results in a specific variance 


tools: 

Boxplot  - Binning - Regression - Clustering 


Business Intelligence

Erstelle und finde Lernmaterialien auf StudySmarter.

Greife kostenlos auf tausende geteilte Karteikarten, Zusammenfassungen, Altklausuren und mehr zu.

Jetzt loslegen

Das sind die beliebtesten StudySmarter Kurse für deinen Studiengang Business Intelligence an der Universität Erlangen-Nürnberg

Für deinen Studiengang Business Intelligence an der Universität Erlangen-Nürnberg gibt es bereits viele Kurse, die von deinen Kommilitonen auf StudySmarter erstellt wurden. Karteikarten, Zusammenfassungen, Altklausuren, Übungsaufgaben und mehr warten auf dich!

Das sind die beliebtesten Business Intelligence Kurse im gesamten StudySmarter Universum

32711 Business Intelligence

FernUniversität in Hagen

Zum Kurs
Business Intelligence & Reporting

TU München

Zum Kurs
Intelligenz

Universität Marburg

Zum Kurs
Business Info

Technische Hochschule Ingolstadt

Zum Kurs
32711 Business Intelligence

FernUniversität in Hagen

Zum Kurs

Die all-in-one Lernapp für Studierende

Greife auf Millionen geteilter Lernmaterialien der StudySmarter Community zu
Kostenlos anmelden Business Intelligence
Erstelle Karteikarten und Zusammenfassungen mit den StudySmarter Tools
Kostenlos loslegen Business Intelligence