predictionio-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Pat Ferrel <>
Subject Re: beginner to ML
Date Sun, 25 Dec 2016 18:03:23 GMT
As a beginner you have chosen a multi-layer problem, each of which would be a good beginner

I’ve found it useful to think of ML in 3 broad categories, classifiers, recommenders, and
regressions. Event NLP is a classifier, taking some words and classifying each.

As Gustav mentioned there are several problems here and in come cases multiple ways to approach
them. So some questions:

1) does your experiment require natural language input?
2) do you want to search for content that is similar to the input or do you want a recommendation
3) probably others

The most simple approach would be a search engine, which would take the natural language input
and find the object with the most words in common. This would probably not be the best you
could do but would be an easy first step. There is a great deal to know about how search engines
work since they are basically a top k sum-of-dot-products similarity engine. This particular
technique is called cosine similarity and is used in many ML aglos.

For this PIO is not needed though you could use it if you wanted to also learn the framework
since PIO includes Elasticsearch.

On Dec 22, 2016, at 10:38 PM, Seshachalam M <> wrote:

Hi All,

I scraped lot of data from a food delivery site(restaurants, menu, prices, user reviews and
ratings). I want to train an ML with this data and able to ask questions like 

"where can i get best hot chocolate near me ?" 
"need biryani for 2 people and a salad, show me restaurants"

It should respond after considering the user reviews/prices/ratings.

I have gone through the prediction io docs but I dont have clear idea on what kind of engines
to use to train with this data.

Please guide me on how to proceed with this.


View raw message