ContinuumIO / anaconda-package-data

Conda package download data
Creative Commons Attribution 4.0 International
98 stars 36 forks source link

Anything wrong with the data from March to today ? #52

Open phwuil opened 5 months ago

phwuil commented 5 months ago

Description

Using condastats, the data show an exponential increase in downloads over the last few months. While we're confident in the quality of our package ;-), this seems unrealistic and, in any case, unexpected (*100 between 2023/12 and 2024/05 !).

Do you have any idea why these variations are occurring ?

What I Did

condastats overall pyagrum --monthly

[...]
          2023-08      2484
          2023-09      2433
          2023-10      4560
          2023-11      3154
          2023-12      1114
          2024-01      2829
          2024-02      2812
          2024-03     12573
          2024-04     66098
          2024-05    110944

Thank you for any hints, explanation or information on this subject

(Copy of https://github.com/sophiamyang/condastats/issues/22)

cappadona commented 5 months ago

Hi @phwuil. We started including .conda packages in our download counts beginning with March 2024 data, which is why you're seeing an increase in downloads.

phwuil commented 5 months ago

Hi @cappadona, thank you for this answer.

How should we compare and understand these figures with those given by the conda-forge badge?

https://anaconda.org/conda-forge/pyagrum/badges/downloads.svg

For our package, at the beginning of May, this badge indicated 764K and today it indicates 795K... +31K which is difficult to compare with the 110K proposed by the stats above... ??

wolfv commented 2 months ago

Looks like the latest data is from 2024-06 and nothing after that.

cappadona commented 2 months ago

Looks like the latest data is from 2024-06 and nothing after that.

@wolfv we have temporarily paused publishing new data

cappadona commented 2 months ago

Hi @phwuil. Thanks for pointing this out. I need to verify the functionality of the counter on anaconda.org, but I believe this tracks all-time download counts for a package from the given channel only and the conda-forge/pyagrum count is currently at 997,376.

Screenshot 2024-09-06 at 1 49 38 PM

For our package, at the beginning of May, this badge indicated 764K anaconda.org/conda-forge/pyagrum/badges/downloads.svg

Is there a way to limit the counter to a certain date range or have you just recorded the count for pyagrum at different dates? If it's the latter, are you willing to share some additional data points?

phwuil commented 2 months ago

hi @cappadona, it was just a comparison with a counter's value noted in a mail in April. However, starting this weekend, I made a small script that daily store the anaconda.org's values.

07-09-2024 03:35 999059
08-09-2024 03:35 1000254
09-09-2024 03:35 1002566

I can give you (privately) an URL for this file if you need it (daily updated).

cappadona commented 2 months ago

Understood. Thanks @phwuil. We'll be investigating the anaconda.org download counter this week and I will reach out if we need access to the results of your daily counts 🙏🏼

phwuil commented 1 month ago

Hi, just an update for the values for pyAgrum on anacondas page :

07-09-2024 03:35 999059
08-09-2024 03:35 1000254
09-09-2024 03:35 1002566
10-09-2024 03:35 1009652
11-09-2024 03:35 1013679
12-09-2024 03:35 1015204
13-09-2024 03:35 1016153
14-09-2024 03:35 1017186
15-09-2024 03:35 1018109
16-09-2024 03:35 1019457
17-09-2024 03:35 1020483
18-09-2024 03:35 1021448
19-09-2024 03:35 1022623
20-09-2024 03:35 1023754

I hope you will find the time to fix these stats ...