Creating a tree using json from a list/table

You should create classes to model your data, in the structure you require. You are basically wanting to build a hierarchical structure from some row based data, this is quite like an XML document, which might be an appropriate solution. But you got me hooked so I played about with what I had before and came up with this:

public class Test { 

    public static void main(String[] args) 
    {
        // hierarchical data in a flattened list
        String[][] data = {
                {"Toyota", "Gas", "Compact", "Corolla"},
                {"Toyota", "Gas", "Compact", "Camry"},
                {"Toyota", "Hybrid", "Compact", "Prius"},
                {"Honda", "Gas", "Compact", "Civic"}
        };

        TreeManager treeManager = new TreeManager();

        for(String[] row : data)
        {
            // build the path to our items in the tree
            List<String> path = new ArrayList<String>();
            for(String item : row)
            {
                // add this item to our path
                path.add(item);
                // will add it unless an Item with this name already exists at this path
                treeManager.addData(treeManager, path);
            }
        }

        treeManager.getData(data[0]).putValue("MPG", 38);
        treeManager.getData(data[1]).putValue("MPG", 28);

        Gson gson = new Gson();

        System.out.println(gson.toJson(treeManager));
    }

    /**
     * This base class provides the hierarchical property of
     * an object that contains a Map of child objects of the same type.
     * It also has a field - Name
     *
     */
    public static abstract class TreeItem implements Iterable<TreeItem>{

        private Map<String, TreeItem> children;     
        private String name;

        public TreeItem() {
            children = new HashMap<String, TreeItem>();
        }

        public String getName() {
            return name;
        }

        public void setName(String name) {
            this.name = name;
        }

        public void addChild(String key, TreeItem data) 
        {           
            children.put(key, data);
        }

        public TreeItem getChild(String key) 
        {           
            return children.get(key);
        }

        public boolean hasChild(String key) 
        {           
            return children.containsKey(key);
        }

        @Override
        public Iterator<TreeItem> iterator() {          
            return children.values().iterator();
        }           
    }

    /**
     * This is our special case, root node. It is a TreeItem in itself
     * but contains methods for building and retrieving items from our tree
     *
     */
    public static class TreeManager extends TreeItem
    {       
        /**
         * Will add an Item to the tree at the specified path with the value
         * equal to the last item in the path, unless that Item already exists 
         */
        public void addData(List<String> path)
        {
            addData(this, path);
        }

        private void addData(TreeItem parent, List<String> path)
        {
            // if we're at the end of the path - create a node
            String data = path.get(0);
            if(path.size() == 1)
            {
                // unless there is already a node with this name
                if(!parent.hasChild(data))
                {
                    Group group = new Group();
                    group.setName(data);
                    parent.addChild(data, group);
                }
            }
            else
            {
                // pass the tail of this path down to the next level in the hierarchy
                addData(parent.getChild(data), path.subList(1, path.size()));
            }
        }

        public Group getData(String[] path)
        {
            return (Group) getData(this, Arrays.asList(path));
        }

        public Group getData(List<String> path)
        {
            return (Group) getData(this, path);
        }

        private TreeItem getData(TreeItem parent, List<String> path)
        {
            if(parent == null || path.size() == 0)
            {
                throw new IllegalArgumentException("Invalid path specified in getData, remainder: " 
                        + Arrays.toString(path.toArray()));
            }
            String data = path.get(0);
            if(path.size() == 1)
            {
                return parent.getChild(data);
            }
            else
            {
                // pass the tail of this path down to the next level in the hierarchy
                return getData(parent.getChild(data), path.subList(1, path.size()));
            }
        }
    }

    public static class Group extends TreeItem {

        private Map<String, Object> properties;

        public Object getValue(Object key) {
            return properties.get(key);
        }

        public Object putValue(String key, Object value) {
            return properties.put(key, value);
        }

        public Group () {
            super();
            properties = new HashMap<String, Object>();
        }       
    }
}

I think this meets most of the requirements you've mentioned so far, although I left out the averaging of the MPG values as an exercise for the reader (I've only got so much time...). This solution is very generic - you may want more concrete sub-classes that better describe your data model (like Manufacturer, Type, Model), as you will then be able to hang more useful methods off them (like calculating averages of fields in child objects) , and you wouldn't have to deal with the properties as a collection of Objects, but you then get more complicated code initialising your data structure from the list. Note - this is not production ready code, I've just provided it as an example of how you might go about modelling your data in Java.

If you are new to not only Java but Object Orientated Programming then you should read up on the subject. The code I have written here is not perfect, I can already see ways it could be improved. Learning to write good quality object orientated code takes time and practice. Read up on Design Patterns and Code Smells .