Recherche avancée

Sur d’autres sites (10246)

• How you can use the Piwik AOM plugin to improve your data and make better online marketing decisions

  13 septembre 2017, par André Kolell — Plugins, AdWords, Bing, Case Study, Criteo, eccommerce tracking, Facebook Ads, how to install piwik analytics, how-to piwik, integration, piwik, Taboola

  Hi, this is André, one of the authors of the Piwik Advanced Online Marketing plugin, which has just hit 5,000 downloads on the Piwik marketplace. In this blog post I’ll show you how Piwik AOM improves your data and enables you to make better online marketing decisions.

  Piwik itself is excellent in tracking all kinds of visitor data, like where a visitor is coming from and what he’s doing on your page or app (pageviews, events, conversions). But what Piwik did not yet take a closer a look at, is how much you’ve invested into your marketing activities and how profitable they are.

  With the Piwik AOM plugin you can integrate data like advertising costs, advertising campaign names, ad impressions etc. from advertising platforms (such as Google AdWords, Microsoft Bing, Criteo, Facebook Ads and Taboola) and individual campaigns (such as such as cost per view/click/acquisition and fixed price per months deals) into Piwik and combine that data with individual Piwik visits.

  Piwik AOM adds a new marketing performance report to Piwik giving you a great overview of all your marketing activities with drill-down functionality :

  

   

  When taking a look at a specific visitor, Piwik AOM shows you the exact cost of acquiring a specific visit :

  

   

  Leveraging Piwik AOM’s full potential

  But although you can access Piwik AOM’s valuable data directly in the Piwik UI for ad-hoc analyses, Piwik AOM’s true strength comes into play when working with the raw data in an external business intelligence application of your choice, where you can further integrate Piwik AOM’s data with your most accurate backend data (like conversion’s contribution margins after returns, new vs. existing customer, etc.).

  Piwik AOM offers some API endpoints that allow you to fetch the data you need but you can also retrieve it directly from Piwik AOM’s aom_visits table, which includes all visits, all allocated advertising costs and advertising campaign details. As there is never data being deleted from aom_visits, the table can easily be connected to your ETL tool with its last update timestamp column. A third way to get data out of Piwik AOM is by developing your own Piwik plugin and listening to the AOM.aomVisitAddedOrUpdated event, which is posted whenever an aom_visits record is added or updated.

  Integrating Piwik AOM’s data with your backend data in the business intelligence application of your choice allows you to evaluate the real performance of your online marketing campaigns when applying different conversion attribution models, conduct customer journey analyses, create sophisticated forecasts and whatever you can think of.

  AOM Use case

  A company that followed this approach, is FINANZCHECK.de, one of Germany’s leading loan comparison websites. At the eMetrics summit 2016 in Berlin, Germany, I gave a talk about FINANZCHECK’s architectural online marketing setup. Until recently, FINANZCHECK used Pentaho data integration to integrate data from Piwik, Piwik AOM and additional internal tools like its proprietary CRM software into Jaspersoft, its data warehouse an BI solution. The enriched data in Jaspersoft was not only used for reporting to various stakeholders but also for optimising all kinds of marketing activities (e.g. bids for individual keywords in Google AdWords) and proactive alerting. Not long ago, FINANZCHECK started an initiative to improve its setup even further – I’ll hopefully be able to cover this in a more detailed case study soon.

  Roadmap

  In the past, we had the chance to make great progress in developing this plugin by solving specific requirements of different companies who use Piwik AOM. During the next months, we plan to integrate more advertising platforms, reimplement Facebook Ads, improve the support of individual campaigns and work on the general plugin stability and performance.

  Before you install Piwik AOM

  Before installing Piwik AOM, you should know that its initial setup and even its maintenance can be quite complex. Piwik AOM will heavily modify your Piwik installation and you will only benefit from Piwik AOM if you are willing to invest quite some time into it.

  If you are not familiar with Piwik’s internals, PHP, MySQL, database backups, cronjobs, creating API accounts at the advertising platforms or adding parameters to your advertising campaign’s URLs, you should probably not install it on your own (at least not in your production environment).

  Piwik AOM has successfully been tested with up to 25k visitors a day for a period of more than two years, running on an AWS server with 4 GB RAM, once CPU and a separate AWS RDS MySQL database.

  Ideas and Support

  If you have ideas for new features or need support with your Piwik AOM installation or leveraging your marketing data’s potential in general, feel free to get in touch with the plugin’s co-author Daniel or me. You can find our contact details on the plugin’s website http://www.advanced-online-marketing.com.

  How to get the Piwik AOM plugin ?

  The Piwik AOM plugin is freely available through the Piwik marketplace at https://plugins.piwik.org/AOM

  Did you like this article ? If yes do not hesitate to share it or give your feedback about the topic you would like us to write about.

• AttributeError : 'FFmpegWriter' object has no attribute '_proc'

  27 février 2024, par Arek

  I have some problem which i cannot find answer to. After installing scikit-video and FFmpeg i got this error :

  



  AttributeError : 'FFmpegWriter' object has no attribute '_proc'

  



  can you help me to find solution to this.

  



  from skvideo.io import FFmpegWriter

def main():
    ...
    video_writer = FFmpegWriter('video.mp4')
    ...
if __name__ == '__main__':
    main()


  



  Already tried to install pyaudio, reinstall FFmpeg and skvideo and install different versions of packages. Does not help at all.

  



  Edit : an example of the full traceback resulting from /skvideo/io/abstract.py is below.

  



  Traceback (most recent call last):&#xA;File "run_modules.py", line 93, in <module>&#xA;    Pipeline.create_videos(video_attr, args.output_path, args.padded)&#xA;File "/home/leuko/.local/lib/python3.6/site-packages/skvideo/io/abstract.py", line 474, in close&#xA;    if self._proc is None:  # pragma: no cover&#xA;AttributeError: &#x27;FFmpegWriter&#x27; object has no attribute &#x27;_proc&#x27;&#xA;</module>

  


• Python ImageIO : Too many open files

  24 août 2017, par orbv12

  I am using imageio in python in order to open all video files in a directory and convert them to numpy arrays.

  Here is the script I am using :

    1 from __future__ import print_function
  
    2 from avi_to_numpy import *
  
    3 from os import listdir
  
    4 import numpy as np
  
    5 import imageio 
  
    6       
  
    7 class_path = '../Diving/'
  
    8 max_frames = 16
  
    9 stride = 8
  
   10 videos = [vid for vid in listdir(class_path)]
  
   11 train = []
  
   12 
  
   13 for vid in videos:
  
   14     print(str.format('Loading {}...', vid), end="")
  
   15     filename = class_path + vid
  
   16     reader = imageio.get_reader(filename, 'ffmpeg')
  
   17     frames = []
  
   18     
  
   19     for i, im in enumerate(reader):
  
   20         if len(frames) == max_frames:
  
   21             break
  
   22         
  
   23         if i % stride == 0:
  
   24             frames.append(im)
  
   25     
  
   26     reader.close()
  
   27     train.append(np.array(frames))
  
   28     print('done')        
  
   29 
  
   30 
  
   31 print(len(train))

  Eventually this script crashes with the following error output :

  Traceback (most recent call last):
  
    File "load_class_test.py", line 16, in <module>
  
      reader = imageio.get_reader(filename, 'ffmpeg')
  
    File "/usr/local/lib/python2.7/site-packages/imageio/core/functions.py", line 111, in get_reader
  
      return format.get_reader(request)
  
    File "/usr/local/lib/python2.7/site-packages/imageio/core/format.py", line 158, in get_reader
  
      return self.Reader(self, request)
  
    File "/usr/local/lib/python2.7/site-packages/imageio/core/format.py", line 207, in __init__
  
      self._open(**self.request.kwargs.copy())
  
    File "/usr/local/lib/python2.7/site-packages/imageio/plugins/ffmpeg.py", line 260, in _open
  
      self._initialize()
  
    File "/usr/local/lib/python2.7/site-packages/imageio/plugins/ffmpeg.py", line 326, in _initialize
  
      stdout=sp.PIPE, stderr=sp.PIPE)
  
    File "/usr/local/Cellar/python/2.7.11/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 710, in __init__
  
      errread, errwrite)
  
    File "/usr/local/Cellar/python/2.7.11/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 1223, in _execute_child
  
      errpipe_read, errpipe_write = self.pipe_cloexec()
  
    File "/usr/local/Cellar/python/2.7.11/Frameworks/Python.framework/Versions/2.7/lib/python2.7/subprocess.py", line 1175, in pipe_cloexec
  
      r, w = os.pipe()
  
  OSError: [Errno 24] Too many open files
  
  </module>

  I am closing the Reader object from imageio. It seems as if the files opened by ffmpeg are not being closed properly.

  Is there an obvious step I am missing here ? Am I closing the files properly ?

  EDIT : Found temporary solution. Opened a new issue on github.

  I was able to resolve the issue by uncommenting the following lines of code from ’imageio/plugins/ffmpeg.py’ :

  381         def _close_streams(self):
  
  382             for std in (self._proc.stdin,
  
  383                         self._proc.stdout,
  
  384                         self._proc.stderr):
  
  385                 try:
  
  386                     std.close()
  
  387                 except Exception:  # pragma: no cover
  
  388                     pass

  I then added a call to the above function in _close(self) :

  271         def _close(self):
  
  272             self._terminate(0.05)  # Short timeout
  
  273             self._close_streams()
  
  274             self._proc = None

  I am not sure what the side effects of doing this are, but it provides a solution for me.

  Here is the link to the issue : https://github.com/imageio/imageio/issues/145