Detektion af regressioner hos Microsoft

 

 Adam Lass, Jesper Rusbjerg, Nikolai Perlt og Rasmus Helsgaun

Løsningen

DEMO

Løsningen DEMO

Bracketing Algoritme

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

Detected Regression

KS-test

D

Alpha = 0.05

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

Score calculation

Old Runtimes

New Runtimes

Calculate Average Runtime

New Avg > Old Avg

No

return

Yes

Calculate Scores

Old Score = Old Avg * Old RPD

New Score = New Avg * New RPD

SDPD = New Score - Old Score

Calculated Values

Old Score

New Score

SDPD

New Avg

Old Avg

New RPD

Old RPD

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

User Impact

Calculated Values

Old Score

New Score

SDPD

New Avg

Old Avg

Calculations

IQR = (Q3 - Q1 * 1.5) + Q3

Upper Bracket Limit = IQR * Old RPD

New RPD

Old RPD

User Impact

SDPD =

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

System Impact

Calculated Values

Old Score

New Score

SDPD

New Avg

Old Avg

New RPD

Old RPD

User Impact

System Impact

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

Calculated Values

Old Score

New Score

SDPD

New Avg

Old Avg

New RPD

Old RPD

User Impact

System Impact

Bracket

Detected Regression

Bracketing Algoritme

Score calculation

User Impact

System Impact

Bracket

Machine Learning vs.

Statistics

  Natural

  Trend

  Shift

Machine Learning vs. Statistics

Machine Learning

Machine Learning vs. Statistics

Machine Learning

Machine Learning

  • Flexible
  • Multiple Variables

Pros

Cons

Machine Learning vs. Statistics

  • Explainability
  • Required loads of training data

Machine Learning vs. Statistics

Statistics

Control Charts

Machine Learning

Control Charts

  • Flexible
  • Multiple Variables

Pros

Cons

Machine Learning vs. Statistics

  • Explainability
  • Credability
  • Single Variable
  • Not Flexible
  • Explainability
  • Required loads of training data
  • Works out of the box

Proces

Proces

Bøhms & Turners Model

Overblik over proces

Scrum

Praktikforløb

Videreudviklingsforløb

XP

Backlog

Scrum Meetings

- Sprint Planning

- Sprint Review

- Retrospective

- Stand-up

Scrum Master

Product Owner

Sprints

Backlog

Scrum Meetings

- Sprint Planning

- Sprint Review

- Retrospective

- Stand-up

Scrum Master

Product Owner

Sprints

Pair Programming

Collective Ownership

Continuous Integration

Refactoring

Planning Game

Pair Programming

Collective Ownership

Continuous Integration

Refactoring

Planning Game

Ikke anvendt

Delvist anvendt

Anvendt

Spørgsmål?

Praktik Hovedopgave - Microsoft

By Adam Lass

Praktik Hovedopgave - Microsoft

  • 299