On Thu, Mar 3, 2016 at 10:43 AM, Blondeau Vincent <vincent.blondeau@worldline.com> wrote:
Hi,
That's a very good idea. For now, I use a Pharo-R binding to make the statistics, but it is better if it is directly in the image.
For my part, it should be nice to have: - Principal Component Analysis - Time series
I already a very simple time series class. What kind of operations you want to do on time series ?
- Correlation with several variables (Correlation matrix) + significance tests (pearson, spearman...) - Standard statistical tests: normality, variance, mean, distribution - And the more important is to have nice tools to visualize the data (Box plot, histogram, scatterplots, PCA ones) with a simple API for standard things but that we can customize. The best is that we are able to draw regression lines on them. But I think that is more a Roassal part.
You have to ask Alex to the viz :-) Please join the SciSmalltalk ml in order to express your needs ;-) -- Serge Stinckwich UCBN & UMI UMMISCO 209 (IRD/UPMC) Every DSL ends up being Smalltalk http://www.doesnotunderstand.org/