#ATAGTR2018
Machine Learning as Decision support
system for QA Professionals
Kaushik Raghavan, Resileo Labs & IITM - Chennai
27th September 2018
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Introduction and Agenda
• Introduction to Machine Learning
• Examples of Machine Leaning
• Regression , Prediction, classification and clustering
• Natural language processing
• Bug count and release date prediction using multiple linear
regression
• Automatic bug classification and clustering
• Generating language agnostic test cases and automatic
requirement mapping
• Conclusion and Future work
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Introduction to Machine Learning
Machine Learning is an idea to learn from examples and experience,
without being explicitly programmed.
Instead of writing code, you feed data to the generic algorithm, and it
builds logic based on the data given.
Machine learning brings together computer science and statistics to
harness that predictive power.
The goal of machine learning is to program computers to use example data
or past experience to solve a given problem.
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
• Examples of Machine Leaning
Smart Email
Categorization
•Spam Filtering
•Primary/Social/
Promotions
Google’s AI-Powered
traffic Predictions
•Goole Maps
•Faster Route Suggest
Ridesharing Apps Like
Ola and Uber
•Uber and Ola Pool
•Minimizing Wait
times
Fraud and Risk
Management in BFSI
•Defaulter Prediction
•Transaction Score
•NPA detection
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
• How Machine learning works
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
• Regression and Prediction
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Regression and Prediction in action
Simple linear regression
Multiple Linear Regression
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Regression and Prediction in action
Simple linear regression
Multiple Linear Regression
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Regression
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Learning Data
Features to predict the bug count
The values to the predicted
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Demo of bug count
prediction using R Studio
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
What else can be predicted
Time to fix a defect (Ridge Regression)
Release dates (Logistic regression)
Bug count trends (ARIMA)
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
• Clustering and classification
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Classification and Clustering
One major area of predictive modeling in data science is classification.
Classification consists of trying to predict which class a particular sample
from a population comes from.
Clustering is a Machine Learning technique that involves the grouping of
data points.
Given a set of data points, we can use a clustering algorithm to classify
each data point into a specific group.
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Automatic Defect Classification using
Naive Bayes Algorithm
The defects can be automatically clustered based on the module they belong to
and severity that needs to be assigned.
There is a lot of manual effort involved towards this task.
This manual effort can be saved by automating the task of classification and
clustering.
Naive Bayes and Decision Trees algorithms were used to create can advanced
classifier.
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Demo of
Automatic Defect classification
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Natural Language Processing
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Generating language agnostic test
cases
For many of us English is not the mother tongue.
Most of us think very well in our mother tongue but fail badly when comes to expressing them
in English.
Most of the time QA teams in India find it difficult to write test cases in English, without
grammatical errors, but they might be very capable of thinking extremely good test cases.
Also team members lack uniformity in the way test cases are generated.
NLP algorithms can come in very handy in this cases.
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Generating language agnostic test
cases
The n- gram model and POS tagger algorithms (which are
well known NLP models) can help build grammatically
correct English sentences from broken English using
sentences as probability models.
Demo: Generating language
agnostic test cases
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Automatic Requirement Mapping to test
cases
One of other main challenge of QA teams face is mapping the requirements to test cases.
This is really a tedious job given the quantum of test cases and requirements for a complex
application.
NLP algorithms liken "TextRank" and "LexRank" can help by automatically suggesting the
requirements to map while writing the test cases.
This way a lot of time is saved by automatically mapping the test cases to requirement.
These ranking algorithms get better over time as scores are assigned to every correct/ wrong
suggestion.
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Automatic Requirement Mapping to test
cases
Demo: Automatic Test case
maping
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Other Use cases
Detecting duplicate defects
Smart Object Identification in test automation using reinforcement
learning.
Server performance degradation using predictive modelling.
Many more….
#ATAGTR2018
As a author of this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media
marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us)
Thank you.
Questions are welcome

#ATAGTR2018 Presentation "Machine Learning as a decision support system for QA professionals" By Kaushik Raghavan

  • 1.
    #ATAGTR2018 Machine Learning asDecision support system for QA Professionals Kaushik Raghavan, Resileo Labs & IITM - Chennai 27th September 2018
  • 2.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Introduction and Agenda • Introduction to Machine Learning • Examples of Machine Leaning • Regression , Prediction, classification and clustering • Natural language processing • Bug count and release date prediction using multiple linear regression • Automatic bug classification and clustering • Generating language agnostic test cases and automatic requirement mapping • Conclusion and Future work
  • 3.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Introduction to Machine Learning Machine Learning is an idea to learn from examples and experience, without being explicitly programmed. Instead of writing code, you feed data to the generic algorithm, and it builds logic based on the data given. Machine learning brings together computer science and statistics to harness that predictive power. The goal of machine learning is to program computers to use example data or past experience to solve a given problem.
  • 4.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) • Examples of Machine Leaning Smart Email Categorization •Spam Filtering •Primary/Social/ Promotions Google’s AI-Powered traffic Predictions •Goole Maps •Faster Route Suggest Ridesharing Apps Like Ola and Uber •Uber and Ola Pool •Minimizing Wait times Fraud and Risk Management in BFSI •Defaulter Prediction •Transaction Score •NPA detection
  • 5.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) • How Machine learning works
  • 6.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) • Regression and Prediction
  • 7.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Regression and Prediction in action Simple linear regression Multiple Linear Regression
  • 8.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Regression and Prediction in action Simple linear regression Multiple Linear Regression
  • 9.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Regression
  • 10.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Learning Data Features to predict the bug count The values to the predicted
  • 11.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Demo of bug count prediction using R Studio
  • 12.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) What else can be predicted Time to fix a defect (Ridge Regression) Release dates (Logistic regression) Bug count trends (ARIMA)
  • 13.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) • Clustering and classification
  • 14.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Classification and Clustering One major area of predictive modeling in data science is classification. Classification consists of trying to predict which class a particular sample from a population comes from. Clustering is a Machine Learning technique that involves the grouping of data points. Given a set of data points, we can use a clustering algorithm to classify each data point into a specific group.
  • 15.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Automatic Defect Classification using Naive Bayes Algorithm The defects can be automatically clustered based on the module they belong to and severity that needs to be assigned. There is a lot of manual effort involved towards this task. This manual effort can be saved by automating the task of classification and clustering. Naive Bayes and Decision Trees algorithms were used to create can advanced classifier.
  • 16.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Demo of Automatic Defect classification
  • 17.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Natural Language Processing
  • 18.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Generating language agnostic test cases For many of us English is not the mother tongue. Most of us think very well in our mother tongue but fail badly when comes to expressing them in English. Most of the time QA teams in India find it difficult to write test cases in English, without grammatical errors, but they might be very capable of thinking extremely good test cases. Also team members lack uniformity in the way test cases are generated. NLP algorithms can come in very handy in this cases.
  • 19.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Generating language agnostic test cases The n- gram model and POS tagger algorithms (which are well known NLP models) can help build grammatically correct English sentences from broken English using sentences as probability models. Demo: Generating language agnostic test cases
  • 20.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Automatic Requirement Mapping to test cases One of other main challenge of QA teams face is mapping the requirements to test cases. This is really a tedious job given the quantum of test cases and requirements for a complex application. NLP algorithms liken "TextRank" and "LexRank" can help by automatically suggesting the requirements to map while writing the test cases. This way a lot of time is saved by automatically mapping the test cases to requirement. These ranking algorithms get better over time as scores are assigned to every correct/ wrong suggestion.
  • 21.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Automatic Requirement Mapping to test cases Demo: Automatic Test case maping
  • 22.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Other Use cases Detecting duplicate defects Smart Object Identification in test automation using reinforcement learning. Server performance degradation using predictive modelling. Many more….
  • 23.
    #ATAGTR2018 As a authorof this presentation I/we own the copyright and confirm the originality of the content. I/we allow Agile testing alliance to use the content for social media marketing, publishing it on ATA Blog or ATA social medial channels(Provided due credit is given to me/us) Thank you. Questions are welcome