Added ownership_transitions derived dataset

datasets/nfts/nfts/cli.py

@@ -18,6 +18,7 @@ from .derive import (
     transfer_statistics_by_address,
     quantile_generating,
     mint_holding_times,
+    ownership_transitions,
     transfer_holding_times,
     transfers_mints_connection_table,
 )
@@ -32,11 +33,12 @@ derive_functions = {
     "current_owners": current_owners,
     "current_market_values": current_market_values,
     "current_values_distribution": current_values_distribution,
-    "transfer_statistics_by_address": transfer_statistics_by_address,
-    "quantile_generating": quantile_generating,
-    "transfers_mints_connection_table": transfers_mints_connection_table,
     "mint_holding_times": mint_holding_times,
+    "ownership_transitions": ownership_transitions,
+    "quantile_generating": quantile_generating,
     "transfer_holding_times": transfer_holding_times,
+    "transfers_mints_connection_table": transfers_mints_connection_table,
+    "transfer_statistics_by_address": transfer_statistics_by_address,
 }
 
 

datasets/nfts/nfts/derive.py

@@ -6,7 +6,6 @@ For example:
 - Current value of each token
 """
 import logging
-from typing import List, Tuple
 import sqlite3
 
 
@@ -109,6 +108,7 @@ def current_owners(conn: sqlite3.Connection) -> None:
         logger.error(e)
 
 
+
 def current_market_values(conn: sqlite3.Connection) -> None:
     """
     Requires a connection to a dataset in which the raw data (esp. transfers) has already been
@@ -456,3 +456,39 @@ def transfer_holding_times(conn: sqlite3.Connection):
         conn.rollback()
         logger.error("Could not create derived dataset: transfer_holding_times")
         logger.error(e)
+
+def ownership_transitions(conn: sqlite3.Connection) -> None:
+    """
+    Derives a table called ownership_transitions which counts the number of transitions in ownership
+    from address A to address B for each pair of addresses (A, B) for which there was at least
+    one transfer from A to B.
+
+    Requires the following tables:
+    - transfers
+    - current_owners
+    """
+    table_name = "ownership_transitions"
+    drop_ownership_transitions = f"DROP TABLE IF EXISTS {table_name};"
+    # TODO(zomglings): Adding transaction_value below causes integer overflow. Might be worth trying MEAN instead of SUM for value transferred.
+    create_ownership_transitions = f"""
+CREATE TABLE {table_name} AS
+WITH transitions(from_address, to_address, transition) AS (
+    SELECT current_owners.owner as from_address, current_owners.owner as to_address, 1 as transition FROM current_owners
+    UNION ALL
+    SELECT transfers.from_address as from_address, transfers.to_address as to_address, 1 as transition FROM transfers
+)
+SELECT
+    transitions.from_address,
+    transitions.to_address,
+    sum(transitions.transition) as num_transitions
+FROM transitions GROUP BY transitions.from_address, transitions.to_address;
+"""
+    cur = conn.cursor()
+    try:
+        cur.execute(drop_ownership_transitions)
+        cur.execute(create_ownership_transitions)
+        conn.commit()
+    except Exception as e:
+        conn.rollback()
+        logger.error(f"Could not create derived dataset: {table_name}")
+        logger.error(e)