Explore
For Enterprise
Join for Free
Log In

artes y humanidades
Negocios
Ciencias de la Computación
Ciencia de Datos
Tecnología de información
salud
Matemáticas y Lógica
Desarrollo Personal
Ciencias Físicas e Ingeniería
Ciencias sociales
Aprendizaje de un idioma
Grados
Certificados

Explorar todo lo que Coursera tiene para ofrecer

Explorar
Buscar
For Enterprise
Inicia Sesión
Únete de forma gratuita

Communicating Results

Loading...

Reproducible Research

Universidad Johns Hopkins

4.5 (3,334 calificaciones) | 52K estudiantes inscritos

Curso 5 de 10 en Ciencia de Datos Programa Especializado

This course focuses on the concepts and tools behind reporting modern data analyses in a reproducible manner. Reproducible research is the idea that data analyses, and more generally, scientific claims, are published with their data and software code so that others may verify the findings and build upon them. The need for reproducibility is increasing dramatically as data analyses become more complex, involving larger datasets and more sophisticated computations. Reproducibility allows for people to focus on the actual content of a data analysis, rather than on superficial details reported in a written summary. In addition, reproducibility makes an analysis more useful to others because the data and code that actually conducted the analysis are available. This course will focus on literate statistical analysis tools which allow one to publish data analyses in a single document that allows others to easily execute the same analysis to obtain the same results.

Destrezas que aprenderás

Knitr, Data Analysis, R Programming, Markup Language

Revisiones

4.5 (3,334 calificaciones)

5 stars
2,213 ratings
4 stars
814 ratings
3 stars
211 ratings
2 stars
58 ratings
1 star
38 ratings

JT

Jul 23, 2017

should be included much more in course 1. it would have been great to know up front how easy it is to mix text and code, not from a reproducibility standpoint, but just to take notes.

IM

Aug 10, 2019

Without taking this course wouldn't have fully understood the importance of reproducible research in data science. Thank you so much. I recommend this course for all data scientists.

De la lección

Week 3: Reproducible Research Checklist & Evidence-based Data Analysis

This week covers what one could call a basic check list for ensuring that a data analysis is reproducible. While it's not absolutely sufficient to follow the check list, it provides a necessary minimum standard that would be applicable to almost any area of analysis.

Communicating Results6:54

Reproducible Research Checklist (part 1)8:21

Reproducible Research Checklist (part 2) 10:20

Reproducible Research Checklist (part 3) 6:54

Evidence-based Data Analysis (part 1)3:51

Evidence-based Data Analysis (part 2) 3:34

Evidence-based Data Analysis (part 3) 4:25

Evidence-based Data Analysis (part 4) 4:47

Evidence-based Data Analysis (part 5) 7:56

Impartido por:

Roger D. Peng, PhD
Associate Professor, Biostatistics
Jeff Leek, PhD
Associate Professor, Biostatistics
Brian Caffo, PhD
Professor, Biostatistics

Transcripción

This lecture is not so much about reproducible research, per se.

But it's more about how you can communicate your results, in a way

that people can understand them, and can check your results, if they need to.

So particularly, when you're doing a large data

analysis, or, you know, a complex kind of project.

And you do your work in a reproducible manner.

You're going to generate a lot of content and a lot of files and a lot of

output and a lot of data. And those all cannot be presented at once.

And so when you present your, your kind of findings to another person you're going to

have, there's going to, there's a kind of

hierarchy of information that you're going to want to present.

Ranging from kind of least specific to most specific.

So the bottom line is that, you know, most people are, these days are very busy.

People like managers and

organization leaders are going to be very busy.

And so they are going to have to kind of accept content that you present.

At a, at a kind of a, at a, at a, at a, in a kind of hierarchical matter.

So the idea is that, you know, many times, when you, when you present a

data analysis, it will be in an oral form like a presentation or in a meeting.

But often the kind of early results or intermediate

results are, are presented via, via a medium like email.

And it's useful to break

down the results of an analysis into

kind of different levels of specificity or granularity.

And so here's a link to just some general advice

on kind of how to get email responses from busy people.

But the basic idea is that you don't want

to kind of blast a ton of information to someone

in one go, because it'll be overwhelming, and people will

be less likely to read what you have to say.

So if you think about you know, what a typical research

paper might look like, you know, there's a, it's

a hierarchy of information built into a research paper.

So typically the first level is going to be the title.

Which is descriptive of what the paper's

about, hopefully it's interesting, it tells you

a little bit about what's going on, but there's really no detail in a title.

Just the kind of the topic that's covered.

The next will be the abstract, which would be, usually

be a couple hundred words, about what the paper's about.

You know, what motivated the problem

what they did to solve the problem.

What the, kind of the bottom line results were.

So that's only going to be in the abstract.

Then of course you have the paper itself.

You know, the body of the paper will show you, will have the, you know, the methods.

And more details about what you really did.

There will be the more detailed results any sensitivity analyses you

know, things can, things like that will be in the paper.

And a much longer discussion of the implications of the results.

But of course,

even the written paper these days, for

a complex data analysis, doesn't specify, really, all

the details that need to be used to need, that are needed to reproduce the findings.

So then you typically might have some supplementary de, materials which

have a lot more of the details about what was done.

And, and then if you really, really want to

kind of do what was done to a level of

precision that is kind of useful, you probably need

to get the code, the data and all the

gory details of kind of what happened.

So that's kind of like the, the range of things that you

might be able to present from least specific to most specific, so

not everyone, of course, is going to be writing a research paper,

but there is an analog to kind of most presentations of data analysis.

And so, if you're going to be you know, mailing results to a person.

Either a colleague, or a manager.

then, you know, the, the first line of information is the

subject of the email, all right, so the subject of the

e-mail is kind of like the title.

And you want to have it, kind of concise, you want to

be descriptive, and at a minimum, you want to have one, all right?

So don't leave it, don't, don't send an e-mail without a

subject, so it's not specific as to what the e-mail is about.

If you can summarize kind of what you've done in one

sentence, the, then it may be useful put that in the subject.

That way people reading the subject can kind of get a sense

of what's happening and maybe even make a decision based on that alone.

The next level of information of course is the email body.

And you, you don't want to go, even though there's no technical limit

on the size of the email body, you don't want to go too crazy

with this, But you want to have, you know provide a brief description of

the problem maybe if someone doesn't, you know, if the person may not remember.

If they're working on many different things at

once, they may not remember what the problem

is, precisely what you're working on, so give

them a little context, give them a little description.

Recall, you know, if there was a meeting previously.

You know, talk about what was

proposed and, you know, and what you actually did.

Summarize some of your findings and your results and maybe, you know, for

a total for one to two paragraphs in this whole, the email body.

If you need this, the person that you're presenting the

information to to take some sort of action based on these,

the results of this presentation, then you should try to

suggest some options that and make them as concrete as possible.

Alright. So and

if there are questions that need to be addressed,

if you want them to kind of comeback with

an answer, it's usually best to try to make

them yes no questions or as simple as possible.

After the email body, of course, you might be able

to attach something that's quite a bit longer, a PDF file

or another type of document, and so this can be like

a report containing more detailed

analyses, figures, tables, things like that.

You can, this can be derived from an R markdown file.

You can use it, you know, something that you create with knitr.

But of course, even in the report like this where you may

be allowed, you know, a couple of pages to, to present things,

you still want to stay concise, you don't want to spit

out pages and pages of code and tables and results.

We know that your code is available because if you

use something like knitr then obviously, you have to kind

of put the code in with the results, but you

don't necessarily have to present it all in the report.

Your opportunity to do something like that of course

will be, if, if you, if you have someone who

really wants to look at what you put, what you

did precisely, you can give em a link to some

sort of repository, like GitHub, or a project

website that would have all of the details.

Or all your code files.

All your data sets.

All your, all that kind of, all the nitty gritty like that.

All your, your software environment, things like that.

And so.

So that's kind of the level of, different levels

of detail that you might want to present to someone.

And so people who are truly interested in everything

that you did, you know, or really want to know

kind of the details, they might go to your

GitHub repository and start pulling your code and start looking

at it and start looking at the detailed results.

But someone who wants kind of top line

or I should say, you know, top line summaries.

You know, they'll, they'll read the email, they'll

look at the subject, they'll read the brief description.

And might flip through the report.

And so different, the point is that different people are going to

be interested in different levels of details of, about what you've done.

So you want to be, and so you want to

present people with those different levels, so that they might

be able to choose the level that they are most interested in.

And so this is just kind of a generic template for kind of how you

might present a an analysis or a data, or a project that you've worked on.

Not every presentation will kind of need all

these different levels but I think that, I

find this to be a useful breakdown of

the different types of presentations that you can make.

Explora nuestro catálogo

Inscríbete de manera gratuita y obtén recomendaciones personalizadas, actualizaciones y ofertas.

Coursera brinda acceso universal a la mejor educación del mundo, al asociarse con las mejores universidades y organizaciones, para ofrecer cursos en línea.

© 2019 Coursera Inc. Todos los derechos reservados.

Descargar en la App Store

Consíguelo en Google Play

Coursera
Acerca de
Liderazgo
Empleo
Catálogo
Certificados
Certificados MasterTrack™
Grados
For Enterprise
For Government

Comunidad
Learners
Partners
Developers
Beta Testers
Translators

Conectar
Blog
Facebook
LinkedIn
Twitter
Instagram
Youtube
Blog de Tecnología

Más
Términos
Privacidad
Ayuda
Accesibilidad
Prensa
Contacto
Directorio
Afiliados