Re: installing environment (stops when compiling "compiler-interface" for Scala)

Memory depends on your data and the engine you are using. Spark puts all data into memory across the Spark cluster so if that is one machine, 4g will not allow more than toy or example data. Remember that PIO and Machine Learning in general works best with big data. BTW my laptop has 16g and I

Upgrading to PredictionIO 0.12.0

PIO-0.12.0 by default, compiles and runs expecting ES5. If you are upgrading (not installing from clean) you will have an issue because ES1 indexes are not upgradable in any simple way. The simplest way to upgrade to pio-0.12.0 and ES5 is to do `pio export` to backup BEFORE upgrading—so export w

Re: [ERROR] [TaskSetManager] Task 2.0 in stage 10.0 had a not serializable result

it work with a very small dataset is because I want to be able to follow the calculations. I want to understand what the UR is doing and understand the impact of changing this or that, here or there... I find that easier to achieve with a small example in which I know exactly what's hap

Re: [ERROR] [TaskSetManager] Task 2.0 in stage 10.0 had a not serializable result

3:1.0})); not retrying [ERROR] [TaskSetManager] Task 2.0 in stage 10.0 (TID 24) had a not serializable result: org.apache.mahout.math.RandomAccessSparseVector Serialization stack: ... Any ideas? On 15 October 2017 at 19:09, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: This is probab

Re: [ERROR] [TaskSetManager] Task 2.0 in stage 10.0 had a not serializable result

"item":"Surface","score":0.0}]} Isn't this odd? Can you guess what's going on? Thank you very much for all your support! noelia On 5 October 2017 at 19:22, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Ok, that config should work. Does the integra

Re: PredictionIO Universal Recommender user rating

Yes, this is a very important point. We have found that the % of video viewed is indeed a very important factor but rather than sending some fraction to indicate the length viewed we have taken the approach before to determine the % that indicates the user liked the video. This we do by trigger

Re: Universal Recommender and PredictionIO 0.12.0 incompatibility

: Hi Pat, On 4 October 2017 at 22:04, Pat Ferrel mailto:p...@actionml.com>> wrote: It looks like PIO 0.12.0 will require a code change in the UR. PIO changed ES1 support drastically when ES5 support was added and broke the UR code. We will do a quick fix to the template to address this.

Re: [ERROR] Timeout when read recent events

When you query for all users in batch, the system is easily overloaded. This is the worst case query situation where no caching applies (for instance). 1) run batch queries at low input load time, because you are competing with input for access to HBase 2) throttle your query speed and/or numbe

Universal Recommender and PredictionIO 0.12.0 incompatibility

It looks like PIO 0.12.0 will require a code change in the UR. PIO changed ES1 support drastically when ES5 support was added and broke the UR code. We will do a quick fix to the template to address this. In the meantime stay on PIO 0.11.0 if you need the UR.

Re: [ERROR] [TaskSetManager] Task 2.0 in stage 10.0 had a not serializable result

What version of Scala. Spark, PIO, and UR are you using? On Oct 4, 2017, at 6:10 AM, Noelia Osés Fernández wrote: Hi all, I'm still trying to create a very simple app to learn to use PredictionIO and still having trouble. I have done pio build no problem. But when I do pio train I get a very

Re: Eventserver API in an Engine?

sion with this thread is to: Enable single-process, single network-listener PredictionIO app deployment (i.e. Queries & Events APIs in the same process.) Attempting to address some previous questions & statements… From Pat Ferrel on Tue, 11 Jul 2017 10:53:48 -0700 (PDT): > h

Re: How to training and deploy on different machine?

e template from machine [TrainingServer] (only need to do this once) Then run `pio deploy` It is not a Spark driver or executor for training Write a cron job of `pio deploy` It is permanent. Thanks Brian On Wed, Sep 20, 2017 at 11:16 PM, Pat Ferrel wrote: > Yes, this is the recom

Re: How to training and deploy on different machine?

Yes, this is the recommended config (Postgres is not, but later). Spark is only needed during training but the `pio train` process creates drives and executors in Spark. The driver will be the `pio train` machine so you must install pio on it. You should have 2 Spark machines at least because th

Re: Unable to connect to all storage backends successfully

meaning is “firstcluster” the cluster name in your Elasticsearch configuration? On Sep 19, 2017, at 8:54 PM, Vaghawan Ojha wrote: I think the problem is with Elasticsearch, are you sure the cluster exists in elasticsearch configuration? On Wed, Sep 20, 2017 at 8:17 AM, Jim Miller mailto:jemi

Re: Universal Recommender - search by subtext/Unicode

the “subtext”? Not sure what you mean by “subtext” - I mean if the I look for an item like "ifone" instead of "iphone", ie. make an error in the spelling, would it still work ?? On Wed, Sep 13, 2017 at 1:37 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: 1) Yes, of

Re: Universal Recommender : seasonality of product

s out of the data"? On Wed, Sep 13, 2017 at 1:42 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: This is done with blacklisting. The default config blacklists all items in the training data that the users has taken the primary event on. So if your primary event is “buy” then

Re: Universal Recommender : seasonality of product

This is done with blacklisting. The default config blacklists all items in the training data that the users has taken the primary event on. So if your primary event is “buy” then once a user has bought a particular table they will not be recommended that table again until the “buy” event ages ou

Re: Universal Recommender - search by subtext/Unicode

1) Yes, of course. Use UTF-8 encoding. 2) I don’t understand this question. The UR is not a search engine, what kind of recommendation are looking for? The best recommendation from a list of items? The best recommendation that contains some text in the “subtext”? Not sure what you mean by “subt

Re: How to give bias on newer item?

With the Universal Recommender there are a robust set of business rules including boosting by item properties. If you add a year+week to every item, then in the query you boost items from a range of weeks by differing amounts you can implement any “decay function you want without modifying code.

Re: Validate the built model

We do cross-validation tests to see how well the model predicts actual behavior. As to the best data mix, cross-validation works with any engine tuning or data input. Typically this requires re-traiing between test runs so make sure you use exatly the same training/test split. If you want to exa

Re: Train a model without stopping

The UR does this automatically. Once deployed you never have to deploy a second time. When a new `pio train` happens the new model is hot-swapped to replace the old, which is then erased, so there is no re-deploy and no downtime. Yes, it uses Elasticsearch aliases but most other Templates do not

Re: Recommender for social media

Actually IMO it is not more complex, it is just far better documented and more flexible. If you don’t need the features it is just as simple as the Apache PIO Templates. I could argue the UR is simpler since you don’t need to $set every item and user, they are determined automatically from the d

Re: Error while running pio train + Universal Recommender

Those are just the Mahout code checks for GPUs. It falls back to using CPUs when they are not found so no error and nothing to fix. On Sep 3, 2017, at 11:27 PM, Saarthak Chandra wrote: Hi, I face the following error while running `pio train` command: [INFO] [RootSolverFactory$] org.apache.m

Re: Securing Event Server on Heroku?

TLS/SSL is required along with authentication of the HTTPS requests. I’m not familiar with Heroku but the Proxy must authenticate the incoming connections. Nginx has basic auth and is a fast proxy, for instance. A cheap, dirty, and not recommended unless it is your only option, is to set your s

Re: Error: Could not find or load main class org.apache.predictionio.tools.console.Console

Ubuntu 16.04.3 x64 -- Paritosh Piplewar Sent with Airmail On 30 August 2017 at 5:22:43 PM, Pat Ferrel (p...@occamsmachete.com <mailto:p...@occamsmachete.com>) wrote: > Can you explain how you installed and what your problem is? The link below > doesn’t contain much information.

Re: Error: Could not find or load main class org.apache.predictionio.tools.console.Console

Can you explain how you installed and what your problem is? The link below doesn’t contain much information. On Aug 29, 2017, at 9:02 PM, Paritosh Piplewar wrote: Yes I'm running pio from inside bin directory. Sent from my iPhone On 30-Aug-2017, at 3:41 AM, Mars Hall mailto:mars.h...@sales

Re: sbt.ResolveException: unresolved dependency: org.apache.predictionio#pio-build;0.11.0-incubrating

haha yeah and fix the typo too but I think the other instructions apply also. Thanks Yevgeny. On Aug 22, 2017, at 9:44 AM, Yevgeny Khodorkovsky wrote: ...incubRating... :) On Tue, Aug 22, 2017 at 09:40 Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Oh, I see which template no

Re: sbt.ResolveException: unresolved dependency: org.apache.predictionio#pio-build;0.11.0-incubrating

core" % pioVersion % "provided", When you are done and it works, consider submitting a PR to the original repo. On Aug 22, 2017, at 9:22 AM, Pat Ferrel wrote: One error is the version of PIO in your templates build.sbt does not match the one you have installed. But not all tem

Re: sbt.ResolveException: unresolved dependency: org.apache.predictionio#pio-build;0.11.0-incubrating

You template is linking to org.apache.predictionio#pio-build;0.10.0-incubrating, what do you have installed? org.apache.predictionio#pio-build;0.11.0-incubrating? Looks like you have to change your templates build.sbt to link to the artifact you have built. On Aug 22, 2017, at 3:52 AM, Abhima

Re: Getting error in

That template has not been moved to use the new Apache namespace. It is still looking for classes with io.prediciton instead of org.apache.predictionio… You will have to update all the imports and build.sbt to use the apache namespace. This fact is noted in the template gallery entry. On Aug 1

Re: Hi I would like to join the google group

Subscribe here: http://predictionio.incubator.apache.org/support/ The site has lot of answers and docs too. On Aug 11, 2017, at 11:39 AM, Trey Zhong wrote: Hi I would like to join the google group of predictionIO user

Re: How to config a high availability eventserver for PredictionIO ?

A truly HA cluster is often not required depending on what you use it for. Can you share what your application is? The EventServer in my experience (I wrote the pages references below) has never crashed because of input. I think the only crash modes I’ve seen involved disk full on some service

Re: Train and deploy on the fly dynamically

PredictionIO only supports batch training of models. If you want online training that updates models with every new input be aware that DL4J is not fast enough to guarantee what some call “kappa” style online learning and PIO does not directly support it. One hack I have used with another onli

Re: Batch Queries

This is a question for the Apache PredictionIO mailing list. I think there has been work done on implementing batch queries in the next version of PIO. Not sure if it will meet your needs. On Aug 4, 2017, at 2:59 PM, l...@platterz.ca wrote: The article about implementing batch queries to PIO

Re: Error when importing data

splitting this apart for production. On Aug 3, 2017, at 8:32 AM, Pat Ferrel wrote: It should be easy to try a smaller batch of data first since we are just guessing On Aug 2, 2017, at 11:22 PM, Carlos Vidal mailto:carlos.vi...@beeva.com>> wrote: Hello Mahesh, Pat Thanks for your answers.

Re: Error when importing data

It should be easy to try a smaller batch of data first since we are just guessing On Aug 2, 2017, at 11:22 PM, Carlos Vidal wrote: Hello Mahesh, Pat Thanks for your answers. I will try with a bigger EC2 instance. Carlos. 2017-08-02 18:42 GMT+02:00 Pat Ferrel mailto:p...@occamsmachete.com

Re: Error when importing data

Actually memory may be your problem. Mahesh Hegde may be right about trying smaller sets. Since it sounds like you have all services running on one machine, they may be in contention for resources. On Aug 2, 2017, at 9:35 AM, Pat Ferrel wrote: Something is not configured correctly `pio

Re: Error when importing data

Something is not configured correctly `pio import` should work with any size of file but this may be an undersized instance for that much data. Spark needs memory, it keeps all data that it needs for a particular calculation spread across all cluster machines in memory. That includes derived da

Re: Using predictionio with existing templates

Yes, and if you get it running, feel free to submit it back to the gallery by updating the docs in a PR. So others can get the benefit of your work. To update docs fork PredictionIO and edit the livedoc branch to create a PR against the Apache PIO livedoc branch. Then a committer will merge to g

Re: Recommendation not taking new data

Could be several things. Did you delete data and re-run the integration test? If you did this it will wipe your data and re-create handmade data. It’s supposed to be an example and test. It’s an example that must be modified to be useful for your data. To be safe create a new app with new data,

Re: Pio build success with error, pio train is faililng.

ng UR. Once done, I would be submitting a PR on the doc repo. Thank you Vaghawan On Thu, Jul 27, 2017 at 9:45 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Yes, a great article but it and the tapster demo do not use the UR. On Jul 27, 2017, at 4:52 AM, Vaghawan Ojha mailto:vaghawan..

Re: Survival Regression Queries

On the Template Gallery there is a github link, you could try creating an issue there to get the author’s attention. I notice there is also a reference for a blog post too under “support” https://github.com/goliasz/pio-template-sr The Gallery is for

Re: Batch recommender

As it happens I just looked into the concurrency issue. How many connections can be made to the Prediction Server. The answer is that Spray HTTP, the earlier version of what is now merged with akka-http, uses something called tcp back-pressure to optimize the number of connections allowed and th

Re: Batch recommender

That feature requires a template to support the eval template APIs and many template do not including the UR. We do cross-validation tests with an integration test that is supplied with the recommender. I'd like to see the eval stuff removed from PIO. Cross-validation is too different for diffe

Re: Pio build success with error, pio train is faililng.

Yes, a great article but it and the tapster demo do not use the UR. On Jul 27, 2017, at 4:52 AM, Vaghawan Ojha wrote: Cool, Now every working template has engine.json and appName field as far as I know. Great it worked. Thanks Vaghawan On Thu, Jul 27, 2017 at 5:13 PM, Darshan A.N. mailto

Re: Survival Regression Queries

Template are supported by their authors. The ones that were donated to Apache are supported by Apache committers here. Others have support links on their gallery entry. The UR has it’s own Google group as do other ActionML templates but often for a non-Apache non-ActionML template you may need t

Re: Deleting / Reversing events in the event server

Blacklisting items in the query happens at query time, events sent to the EventServer will take effect only after the next training time. So item blacklists seems to be the right approach for your issue. On Jul 26, 2017, at 11:52 PM, Mattz wrote: Yes, I think that's what I will do. Thanks for

Re: Pio build success with error, pio train is faililng.

You should take this to the Apache PredictionIO mailing list, this is to support the ActionML applications that use PIIO. Sign up here: http://predictionio.incubator.apache.org/support/ On Jul 25, 2017, at 1:45 PM, darshanan...@gmail.com wro

Re: Intellij Pio Train

cubating", Further for “Intellij sbt refresh, is another problem altogether”: I instead added the project SBT dependencies to “pio-runtime-jars” and added pio-runtime-jars as the class path. This does not result in dependencies missing every time SBT refreshes.   <>

Re: Intellij Pio Train

1.0/storage%20to%20PredictionIO-0.11.0-incubating.Jar> <> From: Pat Ferrel [mailto:p...@occamsmachete.com] Sent: Thursday, July 20, 2017 5:27 PM To: user@predictionio.incubator.apache.org Cc: user-h...@predictionio.incubator.apache.org Subject: Re: Intellij Pio Train +1 This has been a cons

Re: Intellij Pio Train

+1 This has been a constant problem with PIO due to several non-standard build and execution paths in the code. In this case the version of Elasticsearch you use determines which version of PIO client classes are used. I have given up on using a debugger with PIO after having contributed a fair

Re: UR Query Based on Simulated Events

cases. This way we can avoid polluting the event store with dummy data and have to clean it up later as well. Thanks for the quick response, will look into other means to evaluate potential use cases. On Mon, Jul 17, 2017 at 11:00 AM Pat Ferrel mailto:p...@occamsmachete.com>> wrote: But it

Re: UR Query Based on Simulated Events

But it seems you are “simulating” these items, why? how? The fall back mechanisms for new users are pretty flexible in the UR including custom ranking of items, popular, trending, hot, random, etc. On Jul 17, 2017, at 9:54 AM, Pat Ferrel wrote: Personalization cannot work for users with no

Re: UR Query Based on Simulated Events

Personalization cannot work for users with no behavioral data. Your app should be sending events in real time to the EventServer. That way the events below will already be in HBase. Then you just query with the user-id. Item-sets are meant for shopping carts, not events that you know about but h

Re: Does Universal Recommender need Spark for Serving?

A Spark cluster is only needed for `pio train`. Spark must be installed on the machine that runs `pio deploy` but is only used for local client APIs and never needs to communicate with the cluster. However the last I checked EMR is will not work. EMR was designed for Hadoop Mapreduce and Spark

Re: Eventserver API in an Engine?

arbage output." My concern of embedding event server in engine is - what problem are we solving by providing an illusion that events are only limited for one engine? On Wed, Jul 12, 2017 at 12:11 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: "I think to resolv

Re: Eventserver API in an Engine?

is application specific. For example, one can modify the classification to train a classifier on the same set of data used by recommendation. On Tue, Jul 11, 2017 at 10:31 AM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Understood, you have immediate practical reasons for 1 int

Re: Eventserver API in an Engine?

system. On Jul 11, 2017, at 10:31 AM, Pat Ferrel wrote: Understood, you have immediate practical reasons for 1 integrated deployment with the 2 endpoints. But Apache is a do-ology, meaning those who do something win the argument as long as they have enough consensus. I have enough experience

Re: Eventserver API in an Engine?

Understood, you have immediate practical reasons for 1 integrated deployment with the 2 endpoints. But Apache is a do-ology, meaning those who do something win the argument as long as they have enough consensus. I have enough experience with PIO that I have chosen to fix a lot of issues with the

Re: Eventserver API in an Engine?

EventServer then Engines cannot be Kappa online learners and cannot modify models in realtime, a requirement for some lambda batch learners. On Jul 11, 2017, at 9:18 AM, Pat Ferrel wrote: But this is demonstrably untrue. Try it. Clustering for some templates assumes textual data, others do not

Re: Eventserver API in an Engine?

lication. I can create a neural net based on the product description for NLP purpose. etc On Mon, Jul 10, 2017 at 8:10 AM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Good to know but if there is an event blocker and sniffer then they should be a concern of the Engine. Otherwise

Re: Need help with select and setup template

An ALS based recommender template like the E-Com one will handle ratings, but… Rating are not a good way to make recommendations. Netflix found this out after it popularized the idea and does not use ratings now to recommend. After all, would you rate all the movies you like at 10? Of course not

Re: Eventserver API in an Engine?

SQgrCRGXctAbL7eypxwG0ErHpaBJJym15j5Q%40mail.gmail.com%3E> On Sun, Jul 9, 2017 at 5:28 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: I must disagree here, The Engine should decide the disposition of data, which cannot be left to a generic EventServer. Data is the concern of the

Re: Eventserver API in an Engine?

people to store mutable objects and always overwrite. or use better storage structure to capture the changes of mutable object. On Fri, Jun 30, 2017 at 5:29 AM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Actually I think it’s a great solution. The question about diff

Re: Elastic search with PIO

What Template? On Jul 7, 2017, at 7:38 PM, Saarthak Chandra wrote: Hi, I am trying to use elastic search 5.5.0 with PIO 0.11.0, and am getting the following error - "None of the configured nodes are avai

Re: Eventserver API in an Engine?

for people to store mutable objects and always overwrite. or use better storage structure to capture the changes of mutable object. On Fri, Jun 30, 2017 at 5:29 AM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: Actually I think it’s a great solution. The question

Re: Template Suggestions for Collaborative Recommendation

The only Template designed for mulit-modal input is The Universal Recommender On Jul 9, 2017 6:21 AM, "Humantool Project" wrote: > Hello everyone, > > I'm new to the PredictionIO System and would like to set up a basic > Collabroative Recommendation Engine. I went through the Website and found >

Re: Customers clustering

ster_1, even if it's more interested in category10. Am I wrong? 2017-07-07 17:48 GMT+02:00 Pat Ferrel mailto:p...@occamsmachete.com>>: You'll have to work out the ES query JSON, use arrays of strings un-analysed. ES docs indexed cluster_1: [“category 1”, “category 2”] c

Re: Customers clustering

Vandi wrote: Ok got it. But this way, if a customer bought 8 items from category10, and 1 from category1 and category2 it would rank high for cluster_1, even if it's more interested in category10. Am I wrong? 2017-07-07 17:48 GMT+02:00 Pat Ferrel mailto:p...@occamsmachete.com>>: You

Fwd: PIO UR model issue in training

Question was cross-posted Begin forwarded message: From: Pat Ferrel Subject: Re: PIO UR model issue in training Date: July 7, 2017 at 12:35:45 PM PDT To: namita.sharma...@gmail.com Cc: actionml-user Your memory size and cores seem far too low for the data size. Spark gets its speed from the

Re: Customers clustering

on your own This could be put into a simple template but if you already have user history, it may be overkill. On Jul 6, 2017, at 1:39 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: There are 2 clustering templates but it looks like they both need to be moved from Prediction.io &

Re: Customers clustering

you already have user history, it may be overkill. On Jul 6, 2017, at 1:39 PM, Pat Ferrel wrote: There are 2 clustering templates but it looks like they both need to be moved from Prediction.io <http://prediction.io/> to Apache PIO, which should be easy. See the template gallery here:

Re: Customers clustering

There are 2 clustering templates but it looks like they both need to be moved from Prediction.io to Apache PIO, which should be easy. See the template gallery here: http://predictionio.incubator.apache.org/gallery/template-gallery/

Re: Kafka support as Event Store

mas Pocreau Le 5 juil. 2017 21:36, "Pat Ferrel" mailto:p...@occamsmachete.com>> a écrit : No, we try not to fork :-) But it would be nice as you say. It can be done with a small intermediary app that just reads from a Kafka topic and send events to a localhost EventServer, which wo

Re: Kafka support as Event Store

No, we try not to fork :-) But it would be nice as you say. It can be done with a small intermediary app that just reads from a Kafka topic and send events to a localhost EventServer, which would allow events to be custom extracted from say log files (typical contents of Kafka). We’ve done this

Re: Exception - pio train - Caued by permission denied

-Installation/Permission-denied-user-mapred-access-WRITE-inode-quot-quot-hdfs/td-p/16318 <https://community.cloudera.com/t5/CDH-Manual-Installation/Permission-denied-user-mapred-access-WRITE-inode-quot-quot-hdfs/td-p/16318> On Jul 4, 2017, at 2:31 PM, Pat Ferrel wrote: Sebastian: Try logg

Re: Exception - pio train - Caued by permission denied

o re-install it using the user account you intend to use when running PredictionIO. 2017-07-04 21:54 GMT+04:00 Pat Ferrel mailto:p...@occamsmachete.com>>: I you are using the AWS AMI from ActionML you should grant access to user “aml” since PIO is setup to run with this user. On Jul 4, 201

Re: Exception - pio train - Caued by permission denied

I you are using the AWS AMI from ActionML you should grant access to user “aml” since PIO is setup to run with this user. On Jul 4, 2017, at 9:27 AM, Dan Guja wrote: Hi Did you try to give write access to /tmp/ for user ubuntu (as the error says) ? On Tue, Jul 4, 2017 at 10:37 AM, Sebastian

Re: Eventserver API in an Engine?

some committer help. On Jun 30, 2017, at 5:29 AM, Pat Ferrel wrote: Actually I think it’s a great solution. The question about different storage config (https://issues.apache.org/jira/browse/PIO-96 <https://issues.apache.org/jira/browse/PIO-96>) is because Elasticsearch performs the l

Re: Eventserver API in an Engine?

with Elasticsearch meta plus Postgres event & model storage. ``` why user want to use different storage config for different engine? can the classifier match the same configuration as universal recommender? because i thought the storage configuration is more tied to PIO as a whole rather tha

Re: Getting the probability values in the random forest classification template of PredictionIO

You probably should ask on PredictionIO mailing list. Signup here: http://predictionio.incubator.apache.org/support/ On Jun 29, 2017, at 10:33 PM, rbagr...@gmail.com wrote: Hey, We are using the classification template

Re: Eventserver API in an Engine?

.. <> ) > On Jun 28, 2017, at 18:01, Pat Ferrel wrote: > > Ah, one of my favorite subjects. > > I’m working on a prototype server that handles online learning as well as > Lambda style. There is only one server with everything going through REST. > There are 2 resourc

Re: Eventserver API in an Engine?

Ah, one of my favorite subjects. I’m working on a prototype server that handles online learning as well as Lambda style. There is only one server with everything going through REST. There are 2 resource types, Engines and Commands. Engines have REST APIs with endpoints for Events and Queries. S

Re: will property change events affect memory requirements?

Sounds like the only way to remove these will be to export, process yourself to remove all events for deleted items and re-import them. On Jun 25, 2017, at 8:08 AM, Pat Ferrel wrote: File a bug on PredictionIO with this info. the db-cleanger template just uses internal PIO APIs. On Jun 25

Re: UR Model Metrics Introspection

If all recs come back score 0 then they are from the most popular items only, not from the Collaborative Filtering algorithm. If you truly mean “all” then you have a problem in config or nearly 0 data. Don’t mess with tuning until you solve this. On Jun 21, 2017, at 2:27 PM, Daniel Gabrieli w

Re: Can I use one PIO engine for several website?

PIO as a software as a service solution. Is that right ? On Jun 20, 2017 10:07 PM, "Pat Ferrel" mailto:p...@occamsmachete.com>> wrote: Not really, the way to do this is have multiple engines on different ports. On Jun 20, 2017, at 10:03 AM, Amir Jebelli mailto:amir.jebe...@gmail.c

Re: Can I use one PIO engine for several website?

Not really, the way to do this is have multiple engines on different ports. On Jun 20, 2017, at 10:03 AM, Amir Jebelli wrote: Can I deploy one pio engine and use it for several website with different recommendation algorithm?

Re: PredictionIO return array of objects

nge, it would be fine to send items' metadata in as part of the event. If you cannot denormalize and your scale is huge, you might want to consider putting metadata in a KV store. On Fri, Jun 16, 2017 at 12:19 PM, Pat Ferrel mailto:p...@occamsmachete.com>> wrote: If you are settin

Re: PredictionIO return array of objects

w down that path. On Fri, Jun 16, 2017 at 11:27 AM Pat Ferrel mailto:p...@occamsmachete.com>> wrote: What template? Generally this requires you take the id and make a query to your catalog DB. On Jun 16, 2017, at 9:50 AM, Cody Kimball mailto:cody-kimb...@pluralsight.com>> wrote

Re: PredictionIO return array of objects

What template? Generally this requires you take the id and make a query to your catalog DB. On Jun 16, 2017, at 9:50 AM, Cody Kimball wrote: Architectural Design Question: I have a model that performs as expected and returns an array of ID's with their associated scores. Now as I am trying t

Re: Passing an array of objects to get predictions?

predict method in the algorithm class to handle them. I believe this is getting asked more that might warrant a framework level support. It would be great help if you could file a feature request on our JIRA. Regards, Donald On Fri, Jun 16, 2017 at 9:04 AM Pat Ferrel mailto:p

Re: Passing an array of objects to get predictions?

Arrays of events need to be posted to POST http://localhost:7070/batch/events.json?accessKey=… Notice the different URL. Look towards the bottom of this page: https://predictionio.incubator.apache.org/datacollection/eventmodel/

Re: Does predictionIO support pyspark

No, PIO is a framework so calls your template code using JVM bindings so writing a Template in a non-JVM language would be so difficult as to be not worth it. However inside your Scala or Java wrapper template you can use any libs in any language you wish. On Jun 15, 2017, at 11:13 PM, Ravi Ki

Re: Update default build targets

. It's already EOM. > We should upgrade it to 1.2 at least. > > 2017-06-08 5:51 GMT+09:00 Pat Ferrel <mailto:p...@occamsmachete.com>>: >> Supporting the latest and requiring them are 2 different things. Requiring >> them (except for ES) means PIO won’t run

Re: Update default build targets

Supporting the latest and requiring them are 2 different things. Requiring them (except for ES) means PIO won’t run unless the clusters for every user are upgraded to match the client because only backward compatibility is supported. Last time I checked if you require HDFS 2.7, PIO won’t run on

Re: externalizing the spark's es.nodes property from the engine.json

PIO version and template? For the UR, put them in the “sparkConf” section of engine.json named “es.nodes”: “list,of,hosts" comma delimited with no spaces. If not the UR, you may not need this param since the ES master distributes queries in the cluster. On Jun 7, 2017, at 8:42 AM, Dan G

Re: Error while pio status

Sorry for the confusion over support, PIO has many components and the docker container you are using is of unknown origin (to me anyway) It seems to have misconfigured something. Please be sure to tell the author or create a PR for it so it can be fixed for other users, it’s one way to pay for f

Re: Error while training : NegativeArraySizeException

changing the first / primary / conversion event in eventNames changes what the algorithm will predict. CCO can predict anything in the data by changing this conversion events to the one you want. However that means that you must have good data for the primary/conversion event. Removing it will

Re: Error while pio status

This group is for support of ActionML projects like the Universal Recommender. Please direct PIO questions to the Apache PIO mailing list. On Jun 7, 2017, at 6:32 AM, hed...@gmail.com wrote: I'm trying to setup PIO 0.11.0 in docker using this all-in-o

GPUs with The Universal Recommender and PIO

Recently I gave a talk about how Apache Mahout is moving to support GPUs at the lowest layer. This means that the generalized linear algebra of Mahout is GPU accelerated no matter the algorithm constructed with it. So Spark MLlib will have to wait for specialized work to get GPUs (IBM is working

  1   2   3   4   >