predictionio-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Pat Ferrel <...@occamsmachete.com>
Subject Re: Need Help setting up Universal Recommended with PIO
Date Thu, 30 Mar 2017 17:50:48 GMT
When pio became part of Apache we (ActionML) merged our fork and ALL previous templates were
obsoleted because they require code from the io.prediction namespace, instead of the new org.apache.predictionio
namespace.

This means that you should install PredictionIO as described on http://predictionio.incubator.apache.org/
<http://predictionio.incubator.apache.org/> and look in the Template Gallery there for
templates that are Apache PIO compatible, some are still not updated. The UR is there and
points to the correct repo on github.

PredictionIO can be put anywhere and a template should be in another location. I generally
put pio in ~/pio since I build intermediate versions often. I put the UR in ~/ur.



On Mar 30, 2017, at 5:21 AM, Vaghawan Ojha <vaghawan781@gmail.com> wrote:

Hi Thank yo very much for making me clear. 


Thank you very much


On Thu, Mar 30, 2017 at 6:04 PM, Marius Rabenarivo <mariusrabenarivo@gmail.com <mailto:mariusrabenarivo@gmail.com>>
wrote:
You have to run 
./make-distribution.sh

inside incubator-predictionio (the directory containing the repository where you cloned PredictionIO)

it will generate a 
PredictionIO-0.10.0-incubating.tar.gz

that you will have to untar somewhere.
For my case I make the command

tar xvzf PredictionIO-0.10.0-incubating.tar.gz -C ~/

to untar it to my home directory

After that you have to add 
PredictionIO-0.10.0-incubating/bin to you PATH environment variable to be able to run pio
command from any directory.

After that you have to cd to universal-recommender directory

pio build
(this generate the manifest.json)
then
pio train

And

pio deploy

2017-03-30 16:13 GMT+04:00 Vaghawan Ojha <vaghawan781@gmail.com <mailto:vaghawan781@gmail.com>>:
Ok, So universal recommender stays at a different dir independently, so I can't see in the
documentation where it tells me to build pio. Without building how can I deploy. 

Also the quick start tells to pio deploy in the universal recommender dir, in my case it doesn't
work. Since there will be no manifest file. 

Thanks

On Thu, Mar 30, 2017 at 5:53 PM, Marius Rabenarivo <mariusrabenarivo@gmail.com <mailto:mariusrabenarivo@gmail.com>>
wrote:
*universal-recommender

2017-03-30 16:08 GMT+04:00 Marius Rabenarivo <mariusrabenarivo@gmail.com <mailto:mariusrabenarivo@gmail.com>>:
You have just to clone it once and put it in a directory (I put it in my home directory)

You can name it like you want.
When cloning without providing a directory name you will have the repo in univeral-recommender
(It's what I do)

2017-03-30 15:58 GMT+04:00 Vaghawan Ojha <vaghawan781@gmail.com <mailto:vaghawan781@gmail.com>>:
Ok, I am gonna do that now. 

There are lots of confusions in the documentation for me, due to multiple options of installation.
I'm trying a fresh install. 

"Clone Universal Recommender Template from its root repo into ~/universal or do similar for
any other template."

What does this means? am I supposed to clone the UR into pio dir or outside of it? 

And again in the quickstart_doc it says again to clone the UR in root. 

Am I not doing the same thing twice? as per apache encubator, the template should be inside
the pio instllation. 



 

On Thu, Mar 30, 2017 at 5:36 PM, Marius Rabenarivo <mariusrabenarivo@gmail.com <mailto:mariusrabenarivo@gmail.com>>
wrote:
Hi,

As you can see in : https://github.com/actionml/PredictionIO <https://github.com/actionml/PredictionIO>
it was forked from : https://github.com/apache/incubator-predictionio <https://github.com/apache/incubator-predictionio>

What I advise you is cloning the https://github.com/apache/incubator-predictionio <https://github.com/apache/incubator-predictionio>
repo and
git checkout release/0.10.0

2017-03-30 15:43 GMT+04:00 Vaghawan Ojha <vaghawan781@gmail.com <mailto:vaghawan781@gmail.com>>:
Hi,

Are you telling me to use either one of them? 

So let's say I only cloned the first one, the merged branch from actionml. Am I supposed to
skip the later one? like cloning apacheecubator and so on?

Why the documenation is like that? Since you've already installed it, can you please tell
me which could be the best approach? 

Like installing encubator or action ml's? The documentation confused me badly. Please suggest
me. 

On Thu, Mar 30, 2017 at 5:20 PM, Marius Rabenarivo <mariusrabenarivo@gmail.com <mailto:mariusrabenarivo@gmail.com>>
wrote:
Hello,

The first is the ActionML's fork but it was merged with the upstream incubator predictionIO
repo.
You just have to choose on of them.

I think you should use de release branch instead of the master one.

Regards,

Marius

2017-03-30 15:19 GMT+04:00 Vaghawan Ojha <vaghawan781@gmail.com <mailto:vaghawan781@gmail.com>>:
I've ran into confusion as well as after wrong setup probably because I'm not getting this
documentation correctly. 

Documentation URL (http://actionml.com/docs/single_machine <http://actionml.com/docs/single_machine>)

First the documentation tells me to do something like this: 

Clone PredictionIO-0.10.0 from its root repo into ~/pio

git clone https://github.com/actionml/PredictionIO.git <https://github.com/actionml/PredictionIO.git>
pio
cd ~/pio
git checkout master #get the latest stable version


I followed everything in the guide, and the guide again tells me something like this:

Setup PredictionIO


PredictionIO is a source only release so you will need to build it.

Build PredictionIO

git clone https://github.com/apache/incubator-predictionio.git <https://github.com/apache/incubator-predictionio.git>
~/pio
cd ~/pio
git checkout master # usually the version you want
./make-distribution



So, I'm confused why the documentation is telling me something like this, since I already
have the dir called pio. I also tried with the installation into a different dir, and completed
the quickstart of UR, but at the end I just wasted my time. 

So my question here is I can't make two dir with the same name pio. But why the documentation
tells me this? 

If you have any other way of installation please let me know. I would be really grateful,
if you could help me out.

Thank you
Vaghawan














Mime
View raw message