Brief introduction to the use of statistics

[scottdave]

I'm not really sure where to put this. I came across this nice article on Medium. “The 5 Basic Statistics Concepts Data Scientists Need to Know” by George Seif https://link.medium.com/APtnCOapOV

 

[pbuk]

What do you like about that article? I haven't read it all (I gave up at "the first quartile is essentially the 25th percentile") but personally I disagree with the way almost all I have read so far is presented.

 

[scottdave]

Hi @pbuk I would certainly like to know if something I'm reading or Sharing is wrong. It has been years since I've taken a Statistics course, but I am taking a course which uses some. I was looking for something to brush up and came across this. I didn't notice anything glaringly wrong. Please let me know.

 

[WWGD]

What has worked for me is to combine applied and theoretical approaches. I used D.Montgomery's Statistics for Engineers and Schaum's for applied and my old Freund's textbook. Then go over a lot of sites with Stats content.

 

[pbuk]

Its not that I have noticed anything wrong (although as I say I haven't read it all), it's just the way the material is presented - for instance "[the] Median is used over the mean since it is more robust to outlier values". When is the median used in preference over the mean? What even is the mean? Or the median (it is explained as "the line in the middle!")? What does the symbol "IQR" on the chart mean (it is "inter-quartile range of course, and the author talks about the box-plot being "short" or "tall" (in relation to what?) without referring to this label). And then the wonderful tautology I have already quoted - "the first quartile is essentially the 25th percentile". If I don't know what a quartile is how on Earth am I going to know what a percentile is?

When I am looking for an introductory or refresher text, a key indicator is how well it explains the things I already know. If it does a good job, I am inclined to trust the author to explain new material. If it doesn't, then I turn elsewhere.

 

[scottdave]

Its not that I have noticed anything wrong (although as I say I haven't read it all), it's just the way the material is presented - for instance "[the] Median is used over the mean since it is more robust to outlier values". When is the median used in preference over the mean? What even is the mean? Or the median (it is explained as "the line in the middle!")? What does the symbol "IQR" on the chart mean (it is "inter-quartile range of course, and the author talks about the box-plot being "short" or "tall" (in relation to what?) without referring to this label). And then the wonderful tautology I have already quoted - "the first quartile is essentially the 25th percentile". If I don't know what a quartile is how on Earth am I going to know what a percentile is?

When I am looking for an introductory or refresher text, a key indicator is how well it explains the things I already know. If it does a good job, I am inclined to trust the author to explain new material. If it doesn't, then I turn elsewhere.

That makes sense. Thanks.

 

[lomidrevo]

Having just very quick view on the article, I found it quite vague. For example:

A Poisson Distribution is similar to the Normal but with an added factor of skewness.

This doesn't explain anything about Poisson distribution. And no single mention that it is a discrete distribution comparing to Normal, which is continuous.

 

[scottdave]

Having just very quick view on the article, I found it quite vague. For example:

This doesn't explain anything about Poisson...

After rereading it, I agree. It seems like "here are some things to go learn more about."

In doing some searching I came across this site http://rosalind.info It has some problems to try to solve, related to bioinformatics, many of which use statistics concepts. To solve some, writing a program is helpful, as they give a 5 minute time limit, then the data changes. It kind of reminds me of Project Euler style, in a way.