Practical and Modern A/B Test Course – Methods for Data Scientists
A/B Test methodology is widely used in commercial Web Pages, Applications, UI Design and scientific experiments. Optimizing the process and answering business-oriented questions can improve the performance of many goal oriented tasks.
A/B Test Course
This course is about learning the concept of A/B Test.
The course will use Jupyter Notebooks with Python / Julia code within the class in real time.
How do you perform an AB test in data science?
Through instructor-led discussion and interactive, hands-on exercises, participants will understand concept of A/B Test and learn topics such as:
- To introduce the 2 main approaches for A/B Test.
- The multi variants test (A/B/C).
- The Explore vs. Exploit dilemma and few policies to handle it.
- Real world tricks for efficient and reliable tests.
The course will use Jupyter Notebooks with Python / Julia code within the class in real time.
- Data scientists.
- Software engineers.
- Algorithms engineers.
- Basic Python / Julia.
- Basic Probability / Statistics.
- A/B Test
- Definition
- Key Performance Indicators: Discrete (Sales / Clicks) vs. Continuous (Revenue, Revenue per View / Click)
- Frequentist A/B Approach
- The Model
- The T-Test (Continuous and Discrete)
- Estimating number of samples
- Bayesian A/B Approach
- Bayesian vs. Frequentist
- Priors for Bayesian Model and Conjugate Priors
- Discrete Bayesian Model
- Continuous Bayesian Model
- Stopping Rules, Confidence Intervals
- Real World Practices
- Learning prior
- Control group
- Answering business questions by simulations.
- Online Test
- Visualizations
- Introduction to Probabilistic Programming
- Concept
- Frameworks
- Python: PyMC, Pyro
- Julia: Turing
- Modeling: Regression, Robust Regression, Best Tennis Player
- Multi Armed Bandits as Online Multivariate A/B Test
- Definition of the Problem
- Intuition and Basic Policies
- Thomas Sampling
- Reinforcement Learning approach to Multi Armed Bandits
רועי הינו בוגר MSc בהנדסת חשמל ומחשבים מהטכניון. מתמחה בעיבוד אותות, עיבוד תמונה, ראיה ממוחשבת, למידת מכונה (למידה עמוקה) שערוך ואופטימיזציה. רועי מדורג מקום 2 בעולם בקהילת Signal Processing StackExchnage (ה- StackOverflow של עולם עיבוד אותות ותמונה).
- טרם נקבע מועד פתיחה
- 09:00-16:30ימים ושעות
- 24 שעות אקדמיות
- בסיסי ומתקדםרמת הקורס
- עבריתשפת הדרכה
- לבדיקת התאמה לקורס
- ממליצים
- לפרטים אודות קורס אנליסטים מתקדם
- לפתיחה והורדת סילבוס