How DO the Numbers Fool US by Sitara Aghayeva

"Misleading Statistical Facts " ,bizə Data Collection bias, p-dəyərlərinin sirrini və hətta "Black swan or black cat" hadisələrini anlamağımıza kömək edəcək.

Sitarə Ağayeva , Paşa Holdingde Data Scientist və Model Validator expert kimi çalışır.Sitara eyni zamanda data həvəskarlarının bir yerə toplandığı Challengers Deep Educational qurucusu və Women in Data Science icmasının səfirlərindən biridir.

1. HOW DO THE
NUMBERS
FOOL US?
THINGS YOU PREFER NOT
TO THINKABOUT
SITARA AGHAYEVA

2. 01 Modelling the
World
Are all theories true?

3. Modelling is
the way of finding
the nearest
representation of truth.
01
Not the actual truth itself.

4. 02 Do we define
accurately?
Are operational
definitions important?

5. Xoşbaxtlığ, xoşbaxtlığ
02
Not the actual truth itself.
How you define happiness?

6. Examples:
02
But what is happiness?
Imagine reading a research
saying people who eat candies
are happier?

7. 03 Data is source
for knowledge.
But is all data good?

8. Sampling matters.
03
Are we able to reach to
the truth?
Representativeness
Response error
Non-response error
Innate bias

9. Examples:
03
Are we able to reach to
the truth?
Survey about depression
Survey about access to internet
Survey about salary
Crime rates in Malta

10. We see what we want.
03
Are we willing to reach to
the truth?
Confirmation bias
Selection bias

11. Examples:
03
Are we willing to reach to
the truth?
People who leaves universities
become successful, because Bill
Gates did. (btw what is
success?)

12. Why does more
data distort us?
Answer in the last slide.

13. 04 Finding a meaning
in randomness
Is P-value honest?

14. Less probable things are
more probable if you try
enough chances.
04
Monte-Carlo simulation

15. P-value hacking
04
Monte-Carlo simulation
Try enough experiments and
at least one of them will make
you right.

16. 05 Correlation and
causation
Are we meant to be
together?

17. A car that racing you
every day: is s/he really for
you?
05
Ceteris Paribus

18. Solution:
05
Try to go other direction.
Look for ceteris paribus (just
stop the car to see s/he will go
on or not.)
Ceteris Paribus

19. " " Stop for a while to see who is
coming with you and who just
coincides with you.

20. 06 Reporting issue
Have we made a great
progress?

21. I got 10.
06
Actual values are not
comparable.
out of 100.

22. Our company increased
sales by 100%.
06
And percents can be
mind blowing.
From 1 to 2.

23. Solution:
06
Require both actual and
percentage reviews.
Calculate different rations, do
not rely on one for performance.
Comprehensive reporting

24. 07 Predictibility
Randomness drives us
mad.

25. What is randomness and
do computers make it?
07
Roll a dice to know.

26. Answer:
07
But why?
Randomness is the thing that we
cannot see the reason behind and
cannot predict.

27. Why does subjectivity
exist?
07
Is not there only
one truth?

28. Elephant analogy

29. Answer:
07
Why does more data
distort us?
We are not able to see everything.
and partial truth may lead us
false.

30. Examples:
07
Are we willing to reach to
the truth?
Imagine your versions in
different people’s mind: you are
the same person, but everyone
remembers you differently.

31. THANKS
A LOT!
SITARA AGHAYEVA

32. Devil’s Advocate

33. Nassim Nicholas Taleb - “Black Swan”
Jim Al Khalili - “Paradox”
Vaclav Smil - “Numbers don’t lie”
Literature