DataSciFact Profile Banner
Data Science Fact Profile
Data Science Fact

@DataSciFact

Followers
195K
Following
124
Media
569
Statuses
11K

Daily data science tweets from @JohnDCook.

https://t.co/Dh0qdR7w4W
Houston
Joined November 2010
Don't wanna be here? Send us removal request.
@DataSciFact
Data Science Fact
@DataSciFact
1 month
Data science consulting https://t.co/pc4jD9Gntr
2
1
22
@DataSciFact
Data Science Fact
@DataSciFact
18 hours
Making sense of principal component analysis
0
2
14
@L5_fantasylife
ファンタジーライフ公式
@L5_fantasylife
8 hours
“Fantasy Life i” NOW ON SALE! This is your other life! Enjoy an adventure that transcends time and space!
0
1
4
@DataSciFact
Data Science Fact
@DataSciFact
20 hours
A note on Zipf’s law https://t.co/FjGx1UBc46
0
0
15
@JohnDCook
John D. Cook
@JohnDCook
2 days
https://t.co/P9ojVSHHud
2
2
30
@DataSciFact
Data Science Fact
@DataSciFact
22 hours
Pre-registration of clinical trials is associated with fewer positive findings
Tweet card summary image
theincidentaleconomist.com
Recently, Austin and I wrote about problems in the reporting of medical research. We asked, How can we get serious about creating an open, valid, and reliable scientific literature? Among the reforms...
0
0
2
@DataSciFact
Data Science Fact
@DataSciFact
23 hours
New post: Trying to fit exponential data https://t.co/fSSuN1uT0M
johndcook.com
When you're fitting an exponential model to data, you may have the wrong model. Even if you have the right model, your uncertainty grows exponentially with time.
0
0
8
@DataSciFact
Data Science Fact
@DataSciFact
3 days
Trying to fit a logistic curve https://t.co/Twc0iyZjsB
0
2
12
@DataSciFact
Data Science Fact
@DataSciFact
4 days
“Re-examination of the 3/4-law of metabolism” and “Toward a metabolic theory of ecology” https://t.co/WyVEscTqdK
0
0
4
@DataSciFact
Data Science Fact
@DataSciFact
5 days
Brief introduction to fractional factorial designs
johndcook.com
Suppose a black box takes N binary inputs but only M of these inputs matter. A fractional factorial experiment design will require far less than 2^N runs.
0
0
9
@DataSciFact
Data Science Fact
@DataSciFact
5 days
Power-law distributions in empirical data
Tweet card summary image
arxiv.org
Power-law distributions occur in many situations of scientific interest and have significant consequences for our understanding of natural and man-made phenomena. Unfortunately, the detection and...
0
3
18
@JohnDCook
John D. Cook
@JohnDCook
6 days
If you A/B test all of the ingredients of a cake individually, the result will be a terrible cake.
6
13
57
@DataSciFact
Data Science Fact
@DataSciFact
6 days
Extrapolating
Tweet card summary image
xkcd.com
0
1
14
@DataSciFact
Data Science Fact
@DataSciFact
6 days
The only families of probability distributions that admit a sufficient statistic whose dimension remains bounded as the sample size increases are exponential families. -- Pitman–Koopman–Darmois theorem
1
2
6
@DataSciFact
Data Science Fact
@DataSciFact
7 days
The transform y = arcsine(√x) can make proportion data approximately normally distributed.
0
4
20
@DataSciFact
Data Science Fact
@DataSciFact
7 days
Information content of age, birthday, and birth date
johndcook.com
How much information is contained in age, birthday, and birth date? Latanya Sweeney's result that this is enough to identify most people.
0
2
11
@DataSciFact
Data Science Fact
@DataSciFact
8 days
R package for generating synthetic data sets
2
0
14
@DataSciFact
Data Science Fact
@DataSciFact
8 days
One person's meta data is another person's data. It's all data, just different perspectives.
0
1
3
@DataSciFact
Data Science Fact
@DataSciFact
11 days
'Data inherently has all of the foibles of being human. Data is not a magic force in society; it's an extension of us.' -- Mark Hansen
0
5
14
@DataSciFact
Data Science Fact
@DataSciFact
11 days
Bagging
en.wikipedia.org
0
0
6
@DataSciFact
Data Science Fact
@DataSciFact
12 days
A small p-value means the data you saw were unlikely, assuming the null hypothesis AND assuming all your implicit assumptions are true.
1
3
40
@DataSciFact
Data Science Fact
@DataSciFact
12 days
Best fit to three data points
johndcook.com
Find the best fit to three data points, best in the sense of minimizing the maximum error.
1
0
7