feat: features stats

poptimizer/__main__.py

@@ -1,7 +1,7 @@
 import typer
 
 from poptimizer import consts
-from poptimizer.cli import app, portfolio
+from poptimizer.cli import app, feat, portfolio
 
 
 def _main() -> None:
@@ -12,6 +12,7 @@ def _main() -> None:
     )
     cli.command()(app.run)
     cli.command()(portfolio.export)
+    cli.command()(feat.stats)
     cli()
 
 

poptimizer/adapters/mongo.py

@@ -115,6 +115,19 @@ async def get[E: domain.Entity](
 
         return self._create_entity(t_entity, doc)
 
+    async def get_all[E: domain.Entity](
+        self,
+        t_entity: type[E],
+    ) -> AsyncIterator[E]:
+        collection_name = adapter.get_component_name(t_entity)
+        db = self._db[collection_name]
+
+        try:
+            async for doc in db.find({}):
+                yield self._create_entity(t_entity, doc)
+        except PyMongoError as err:
+            raise errors.AdapterError("can't load entities from {collection_name}") from err
+
     async def _load(self, collection_name: str, uid: domain.UID) -> MongoDocument | None:
         collection = self._db[collection_name]
 

poptimizer/cli/feat.py

@@ -0,0 +1,69 @@
+import asyncio
+import contextlib
+import statistics
+from collections import Counter
+
+import uvloop
+
+from poptimizer import config
+from poptimizer.adapters import logger, mongo
+from poptimizer.domain.evolve import evolve
+
+
+async def _run() -> None:
+    cfg = config.Cfg()
+    err: Exception | None = None
+
+    async with contextlib.AsyncExitStack() as stack:
+        lgr = await stack.enter_async_context(logger.init())
+
+        mongo_db = await stack.enter_async_context(mongo.db(cfg.mongo_db_uri, cfg.mongo_db_db))
+        repo = mongo.Repo(mongo_db)
+
+        lgr.info("Starting...")
+
+        try:
+            count = 0
+            history_days: list[int] = []
+            features: Counter[str] = Counter()
+
+            async for model in repo.get_all(evolve.Model):
+                if not model.ver:
+                    continue
+
+                count += 1
+                batch = model.phenotype["batch"]
+                history_days.append(batch["history_days"])
+                features.update({"use_lag_feat": batch["use_lag_feat"]})
+                features.update(batch["num_feats"])
+                features.update(batch["emb_feats"])
+                features.update(batch["emb_seq_feats"])
+
+            lgr.info("Count - %d", count)
+            lgr.info(
+                "History days - %d - %d - %d",
+                min(history_days),
+                statistics.median(history_days),
+                max(history_days),
+            )
+            for feature, feat_count in features.most_common():
+                lgr.info(f"Feature {feature} - {feat_count / count:.2%}")
+
+            lgr.info("Finished")
+        except asyncio.CancelledError:
+            lgr.info("Shutdown finished")
+        except Exception as exc:  # noqa: BLE001
+            lgr.warning("Shutdown abnormally: %r", exc)
+            err = exc
+
+    if err:
+        raise err
+
+
+def stats() -> None:
+    """Features genotype statistics."""
+    uvloop.run(_run())
+
+
+if __name__ == "__main__":
+    stats()

poptimizer/cli/portfolio.py

@@ -24,8 +24,9 @@ async def _run(out: Path) -> None:
         lgr = await stack.enter_async_context(logger.init())
         repo = mongo.Repo(mongo_db)
 
+        lgr.info("Starting...")
+
         try:
-            lgr.info("Starting...")
             port = await repo.get(portfolio.Portfolio)
             positions = {pos.ticker: shares for pos in port.positions if (shares := sum(pos.accounts.values()))}
             out.parent.mkdir(parents=True, exist_ok=True)