toru-takahashi

-- Hive
SELECT user, TD_X_RANK(user) AS node_id, page_id, category, time
FROM
(
  SELECT
    raw_data.user, page_id, category, raw_data.time AS time
  FROM
  (
    SELECT user, page_id, category, time
    FROM raw_data_201501

toru-takahashi

-- Hive --
SELECT
  user,
  TD_FIRST(category, node_id) AS landing_category,
  TD_LAST (category, node_id) AS last_category,
  COLLECT_ALL( category )     AS category_set,
 
  TD_FIRST(page_id, node_id) AS landing_node,
  TD_LAST (page_id, node_id) AS last_node,
  COLLECT_ALL( page_id )     AS node_set,