I need to export an H2o model in Javascript or Excel.
I know how to export models using JAVA, Python and R. I want to do the same for JS and Excel.
An idea could be to load the exported model in mojo to the sever and then use the REST API to perform model evaluation. Unfortunately we cannot succeed in performing the loading (we have opened an issue here regarding this specific problem).
Do you have any idea of how to solve this issue?
Thanks
Here is a pointer to a git repo that has an example of how to host an H2O-3 MOJO in a REST API:
https://github.com/h2oai/app-consumer-loan
Related
I have trained, deployed and evaluated a model successfully on google cloud platform's natural language product AutoML Text & Document Classification. When i see all deployed models, there is no option to export the model, only delete option is available. Along with exporting data, is there a way to export the automl model as well, so that I can run the model on my local or any other machine?
There are videos explaining to export model in auto ml vision product, but I couldnt draw much similarities with natural language model. Thanks
This feature doesn't exist yet for Natural Language product but there is a request for it here:
https://issuetracker.google.com/issues/168860629
I would like to create a sample Form in Kibana (version 6.3) to add data to elasticsearch. Is there an example or link which I can follow it to create my UI in first step and loading data to elasticsearch in second step?
Thanks for advance ^^
Kibana itself is not meant for data manipulation - it is a discoverability and visualisation tool. You can edit saved search objects, but even Kibana itself discourages it. The following is a copy/paste from the Search Objects interface:
Typically objects are only modified via their associated application, which is probably what you should use instead of this screen
My suggestion would be to create a simple web app using something like Flask or Padrino, depending on which language you are most comfortable with. For example, this is how to build a form in Flask. You can use the elasticsearch Python module to record data in the index.
After a lot of researchs, I found this two links 1 & 2 (the best documentation I found) to create a plugin with Forms and interaction with Elasticsearch.
Thanks :)
I'd like to use a part of d3 server-side in a Meteor app. Specifically, d3's CSV utilities. However, the package.js file only provides d3 to clients.
Does it make sense to use the d3 CSV utilities server-side? If yes, what's the canonical way to make this happen?
Crossposted on forums.meteor.com here.
You can make your own package or just download and drop the d3.js into your project in a way that makes it accessible server side, as in don't put it in the client or public directories.
You could give D3-Node a try. https://github.com/bradoyler/d3-node
I'm moving an app from flat file php to codeigniter and I'd like to integrate the two as much as possible before moving over completely. I'm looking to specifically use the input class outside of codeigniter but it looks like I'd also need to use the controller logic (to get access to input segments). Can anyone walk me through using the input library in a flat file php?
I think you're creating too much work and potential problems by trying to do this. You'll be better off if you go directly to CI. Move any standalone functions in your standard PHP files into Helper functions.
If you really want to do this, you could use CI and create controllers/functions for all your files, then in the controller functions, just include() your PHP file and ignore the models and view for now. That way you'll have access to all the CI variables, including the $this->input data.
I'm working on learning node.js as a possible tool for web development. I understand how express.js works, but I'm not sure how to integrate mongoose models, I've tried searching google and here for an answer with no luck. Where should models go? Should they go in app.js, with the rest of the configuration and rorouting information? or should they go in a seperate file and imported like a normal module? Any information and resources would be appreciated. Thank you.
I would recommend putting your models in separate files even if just for code organisation purposes. Create them as normal node modules and export the schema. Import the schema into your app.js and model them there:
var ModelSchema = require('models/model');
var model = mongoose.model('Model', ModelSchema);