Adding a gui to a book and more!

[rdstern]

I would like to explore this idea further with @jkmusyoka and @rachelkg in relation to one, or more, of his courses in 2023. I am keen that our "home team" becomes more serious and therefore uses R-Instat in supporting the teaching of statistics and data science. I suggest my formula fits well with Maseno, and also with James' interests in statistical education! There are some principles here and therefore quite a lot that can be reported - in papers and in meetings for someone who specialises in statistics education.

1. Maseno does assume students all do some computing. This fits well with my proposal that it is ridiculous for students to study statistics without having access and skills in using a computer. This does not imply each student needs their own personal computer. In the 1980s much was done with a single computer, shared among a class and our situation is better than that now. If not, then don't try to teach statistics!
2. It will be nicer for R-Instat when more is available from a mobile phone (plus internet access), but I am not prepared to wait.
3. R is excellent, because it is so good, and it is free and open-source. But it can be tricky to use, hence the R-Instat GUI can make R easily available to many more students.
4. A more recent development is the statistics and data science books that are freely available to read online. That's also important, as it makes the books just as available as the R-software. And that is needed in many countries.
5. But the books have the same problems as R itself, namely they assume you learn R first. So, we need to choose books for the course, and write a gui for the books! So that's my proposal.
6. R-Instat incorporates many R packages and helps to make them easy to use. I would like to consider a book as effectively like just another R package. So, if we add R-Instat to the book, it becomes easier to read/use.
7. Taking the parallel further, when we put a normal R package into R-Instat, we don't incorporate all the functions. You can still use the other ones - if you like - through the script window. Similarly, when we add the R-Instat gui to a book, it doesn't have to be for the whole book. It can be useful for just parts of the book. Other bits of our course might make another book easier instead!
8. This fits well with the way R-Instat could be optional. Those who would prefer to learn R first can still do so, while others can graduate later - or not at all!
9. It is only recently that statistics books, now seem to be available (online) freely. More data science books seem to be available. If we would like to apply our "treatment" to a book that isn't (yet) available, then many authors will make efforts for some countries. I suggest that adding books and R-Instat ease-of-use to courses is another useful dimension.
10. Also our current formula for R-Instat fits well, especially with the new and improving script windows. We may not be the easiest GUI (out of those reviewed), but my recent third tutorial shows we are easy enough, even for those students who will always be frightened of commands. We are just what's needed for those who would like to start without the language, and graduate later. And those who prefer, can use the book just as it was written!
11. What's next? Is there a flaw in my argument above? If not, then I would like to confirm that the Introduction to Data Science book will be one that is recommended for the data science MSc. If so, then two part of that book are on data visualisation and data tidying. Perhaps they can be GUId? We could start with both parts, but I suggest the tidying will be simpler, in that R-Instat is more complete there.
12. We need books on statistical modelling. I wonder which?
13. Once we have tested it, then I suggest our formula might give us some extra (paid?) work. It is surprising how many books credit grants that helped them to write the book. If INNODEMS could have a team that can GUI a book, (or starting with just part of the book) perhaps there could sometimes be a grant for that? Once we have a formula I would quite like INNODEMS or IDEMS to sometimes ask authors what they think of this.
14. With an online book ithe text can just stay the same, but you could click where there are some of the elements of code that have a GUI equivalent. It then takes you perhaps to a place in an appendix where there is the corresponding GUI version.
15. In the early stages this will also help us in R-Instat development. Often, when I try to use R-Instat, I can think of little improvements I would like in the dialogues - and that's why we have the programmers! I suggest we will find that in the visualisation - and that will be useful for R-Instat.
16. I also don't want to omit items that we would like to include in books and courses. One particularly important item is tabulation. This always seems to be omitted in visualisation and I would like to include it in our teaching.

[rdstern]

If you agree with the gui-to-a-book idea above, then that's one approach. That will support our courses when they follow one or more books written by others.
I hope we will want to do more and also cover topics we think are important, and may not be well covered yet.
I remember the training with Parin for u/g agriculture students with the computeraid machines. They had 2 main courses, where - in year 2 we just went as far as descriptive statistics, while year 3 went into modelling.
Modernising that idea our problem-solving in statistics course for AIMS doesn't really have any modelling.
I wonder about a course in "Data Science for Development" that concentrates on data???It would look at different sources of data with many examples from lots of different fields.

1. One source is the lists package and we have a video. This shows some of the wide range of applications. Also that you need to be able to manipulate the data to look at it more clearly. So it is natural to include elements of data tidying.
2. Another source is interesting data from the existing R packages, including sometimes looking at the sources of those data - which gives access to more.
3. Another category is that data to some people is the results from a study to others. Data is sometimes semi-processed or analysed, and then people wish to do more with these data.
4. This can include examples where just the results are the new starting point, or when the primary data are also available. The MICS and other large surveys are examples of these.
5. Moving data to the right level could be worth considering here. This introduces the ideas of surveys at different levels and different levels to process climatic data as well as multiple response and repeated measures data.
6. Data that keeps coming - this includes climatic and many other areas (including some that IDEMS are working in) where ideally the processing also takes account of these new data.
7. Dimensions that are sometimes forgotten in data. I am particularly thinking here of spatial data. Here you often have to forget the time component (if time series data) to map the results. Or you may forget the spatial component if you just analyse the different spatial points together.
8. Combining data sets together. Here I am (of course) particularly interested in combining remote and station data, but here I am particularly interested in the 2 sources - and a problem is that often researchers choose one source, because that's more in their comfort zone.

What have I missed. I suggest there is material here for a course in our data science for development course. And I suggest that many data science courses quickly advance to tools and advanced methods, when simple tools are often most of what's needed.

So, this exploration of data might go as far as the descriptive statistics. But I suggest there is enough material for a follow-up course that covers data tidying and summaries and visualisation and tabulation. If we need more then it could include reporting and archiving.

I don't know how bid a course is, but I would be more concerned from the above that we have too much material for 2 Modules, rather than too little. The first course could perhaps be called The Data. The second could perhaps be Data Processing.

In terms of our basic book - Introduction to Data Science, we should relate the the book, but go further. I suggest the Data module covers the data in the book, and the accompanying R package includes many data sets - we go further in our treatment.
The Processing perhaps covers both the Data Visualisation section and the Data Wrangling section. In the Data Visualisation section it goes further, particularly by covering tabulation - including multiple response sets. I assume it covers the Data Wrangling, but perhaps there may be another module that extends what we do here.

The last part of the book is on Productivity Tools. I now wonder whether that should become another Module in our course, rather than trying to include it here. We could mention them here perhaps so the main Module is Optional. I suggest these 2 modules might be compulsory.

An optional course I would quite like to prepare is then R from R-Instat. This is one of 3 optional courses. One is R, the second is python and this is the third. Probably one is compulsory, but not more than 1 will be credited. If more than 1, then the best mark counts. If that's not possible, then they could all 3 be optional - but you still can't get credit for more than 1.

[rdstern]

I have now had a useful discussion with @jkmusyoka and we propose to use R-Instat on his first-year course, which is on descriptive statistics. This doesn't fit with the books idea, but is where R-Instat is strong, so that's great. Some information about his courses is as follows:
a) There are between 300 and 400 students
b) There is a computer lab, but it was meant for when there were 100 students. So organising practical sessions there is not sensible. It is used for self-study and so may be good to install R-Instat on those machines?
c) About 1/4 to 1/2 students have their own machine. Those that don't, share and James encourages this.
d) This also includes good internet use.
e) Demonstrations in a class with a projector isn't practical. Instead James uses zoom and says this works well.
f) I didn't ask about Moodle, but assume they would register on the course Moodle site?
g) I assume we would want some stack exercises. Is that through Moodle?
h) The course starts in February.
i) I gather James made some use of R-Instat last time, including the R-Instat tutorials - and they are being updated.
j) He will add more details about the course content/date etc to this discussion.