高效的 Java 集合分析来自具有数百万条记录的 CSV 文件的输入

Question

Lets say I have a csv file with stock exchange information in following format: timestamp, name, price, qty, account, buy/sell. This file may have millions of records and represents the trading activity for the day. The file is not sorted and I need to choose the most optimal Java collection for holding this data in order to provide analytics efficiently.

Analytics Eg: 1)Most sold stock 2) Account with max transactions 3) Highest quantity of stock bought in a time range. 4) Top K people with highest transactions.

Basically I will need to sort this list many times based on different fields.

So after a little bit of search I found that a Tree based collection is best for this use case. Like a TreeMap. Is there any other collection which would be better?

Answer 1

TreeSet will be efficient if you want sorted by one parameter. You can

1. Create a class like:

    public class Record {
        Calendar timeStamp;
        String name;
        double price;
        //...
    }

2. Create comparators for each task

3. Create a LinkedList (or other Collection)

    List <Record> records=new LinkedList();

4. Use your comparators

    records.sort(yourComparator1);
    records.sort(yourComparator2);
    records.sort(yourComparator3);