Read from File or Database

We have formulated couple of design specifications for a project and would like to have your opinion on them.

The project typically takes CSV files as input and applies "operations" to it. The CSV file would have tens of thousands of records with 50 to 100 columns. One of the objective of the project is to to extract one/few of the columns from it, apply an equation and create a new column/variable.

Let us assume a CSV file named student.csv with columns: Name, Gender, Age, Dept, etc. has 80,000 records/students. Now we want to make a new column/variable named 'Experience' and the equation to apply is: (Age/2 - 10).

We see two possibilities:

1) Use file read option and store the entire content in a PHP variable, fetch the 'Age' column value and populate 'Experience' for each student.

Cons: If I want to fetch only certain records (Ex: Students aged between 20 and 25) or if I want to fetch only particular columns (Ex: Name and Age), the system will need to fetch the complete file, parse it and generate such a report (this might be practically impossible in case of large datasets)

Doubt: Is there some existing API that would fetch particular set of columns from a file?! (Ex: http://code.google.com/p/php-csv-parser/)

2) When the CSV file is imported, this shall be created as a DB Table (CSV first line is the DB Table Columns). And, newly created variable (Ex: Experience) shall be new column.

I feel that the second option is practically possible, but worried about the number of tables that will be generated! If I upload 200 CSV files, it would generate 200 DB Tables, and each table will have tens of thousands of records!

Please advise with your thoughts.

Thanks!