well, let's be more careful when interpreting these statistics, as they
could originate from automated CI infrastructure and do not necessarily
represent actual users. Also we are just about to prepare the first
Apache SystemDS release (so there is no comparison).
Furthermore, yes in SystemDS some old baggage has been removed, lots of
new functionality added, and some APIs will be rebuild over the next
months and years. So if you want to contribute to the new Python
MLContext feel free to create a design document and an initial PR for
its implementation. Note that the old Python MLContext was a thin
wrapper around our Java MLContext, and rather experimental without any
tests. The old Matrix Python API is already superseded by our new lazily
evaluated Python API. Accordingly, we should not aim to copy the old
code again into the repository as this makes it more difficult to build
the right seamless abstractions.
Regards,
Matthias
On 8/20/2020 6:29 AM, Janardhan wrote:
Hi all,
1. The following are the statistics of pypi downloads with an average of `~
10k ` per month.
These results are that of SystemML 1.2.0.
num_downloadsmonth
6,922 Aug, 2020 (till 20th)
14,864 July, 2020
13,163 June, 2020
6,613 May, 2020
10,331 Apr, 2020
13,483 Mar, 2020
9,150 Feb, 2020
4207 Jan, 2020
A draft document available at [1].
2. Why this info?
This is to emphasize that the usage of systemml is still in production. And
about the removal of some of
its functionality in SystemDS releases.
Also, the SystemDS downloads
num_downloadsmonth
316 Aug, 2020
335 July, 2020
468 June, 2020
280 May, 2020
216 Apr, 2020
411 Mar, 2020
[1]
https://github.com/j143/notebooks/blob/master/Decisions/systemds.md#query-run-on-20th-aug-2020
Thank you,
Janardhan
On Sun, Jun 14, 2020 at 7:30 PM Janardhan <[email protected]> wrote:
Hi all,
The following are the statistics of pypi downloads with a maximum of
`~13500` in March 2020.
These are only that of SystemML 1.2.0.
num_downloadsmonth
5077 202006
6613 202005
10331 202004
13483 202003
9150 202002
4207 202001
3443 201912
A draft document available at
https://github.com/j143/notebooks/blob/master/Decisions/systemds.md
This document also contains some companies using SystemML. :)
Thank you,
Janardhan
On Sun, Aug 11, 2019 at 12:24 PM Janardhan <[email protected]>
wrote:
Hi Everyone,
A. No GitHub activity makes the project unpopular as a general rule.
I believe, assessing the stats would be a great way to assure it is
not the case.
B. IMPORTANT NOTE: Folks at Graz University of Technology
<https://www.tugraz.at> are actively working on a fork of
APACHE SYSTEMML-1.2.0 at https://github.com/tugraz-isds/systemds
C. REQUEST FROM PMC. Can you share the following stats? - thanks
Downloads
----------------
1. Jar files from Maven Central, can be accessed by PMC (I believe, if
not one can contact ASF Infra)
at https://repository.apache.org
2. Release artifacts downloads most probably be with systemml PMC
3. Python artifacts data is already summarized.
Website
-----------
1. Google Analytics - I think we have the website statistics, can they
be shared.
2. Algolia statistics - We haven't released algolia search yet.
(but a little over 10 users/week are hitting the bleeding edge
documentation)
Thank you,
Janardhan
On Sat, Aug 10, 2019 at 8:07 PM Matthias Boehm <[email protected]> wrote:
great - thank you so much for the summary. It would be awesome to get it
for a longer history as well and aggregate it with the release/maven
downloads.
Regards,
Matthias
On 10/08/2019 16:23, Janardhan wrote:
Hi,
The following are the queried SystemML download statistics to
understand
the present user base.
Retrived from `the-psf.pypi.downloads`, processed with BiqQuery.
Month num_downloads
Aug-19 1205
Jul-19 3512
Jun-19 3582
May-19 1586
Apr-19 1076
Mar-19 1029
Feb-19 792
[image: image.png]
Thank you,
Janardhan
