Vijayalakshmi,
No matter what tool you use or the write code yourself, you need to extract the metadata about your CSV files before you can expose through odata. Then you have to provide querying capabilities. Know that CSV files do not define any query capabilities like JPA or JDBC interfaces do. So, when you write the code you need to fill the gap of query capabilities for odata when use plain odata4j framework.
Teiid fills all the above gaps, Teiid provides *a* way to structurally arrange the data the way you want to expose to the client applications, and then provides complete querying capabilities on top that data either using SQL or OData. You do not need to write single line of Java code or generate any web apps. Now compare time spent and managing the above to this option.
As per your installation comments, how long does it take to unzip two zip files? As per the learning curve that comes with any software -:) Look at quick start it shows you how to access CSV files
But if your CSV format is changing often, then there may be some advanced ways you can accomplish that in Teiid through automatic metadata detection and re-deployment, that are beyond scope of this thread, that need good knowledge of Teiid workings.
HTH
Ramesh..