How can I implement machine learning algorithms in a web application?

分享！這是 如何把 Machine learning or Deep Learning 演算法，嵌入在一個網站上面，然後，在後面做Machine Learning or Deep Learning。

以下是步驟:

1. You need a database. Your choice depends on different aspects but most important thing 
 is size and speed of your data. For small sized problems a regular RDBMS will do the job.

2. You need a component to build dynamic HTML pages. A typical web programming 
language like PHP will do that job. Your dynamics HTML component manages 
communication with the database on front end.

3. You need a beautiful and easy to use front-end. The skills required are CSS, Javascript, 
and plain HTML. This component communicates to (and partially is generated by) 
component 2.

4.Final component is your ML engine. You can write it in any language but performance 
and type of application are the most considerations. 

For large distributed applications your choice comes down to Hadoop or Spark  
ecosystems. 

For mid-size data sets you can use Java and C++. If you have a small size data, [R] and 
MATLAB can be used. 

Your ML-engine might communicate with the database directly (usually if it's a large 
application or involves online learning) or might not (if you have another mechanism to 
periodically extract data and update your ML-engine). 

The results of the ML-engine is the feed for your 2nd component engine. Something 
among a typical relational database, a text file, or JSON file is common here.

As you can see different skills are involved for a production type web-based ML application. In enterprise level applications, first component is performed by a data engineer, second one by a software (web) developer, third one by a graphic designer (UI engineer), and last one is the work of a data scientist.