feat: build year day feature

poptimizer/consts.py

@@ -2,7 +2,7 @@
 from pathlib import Path
 from typing import Final
 
-__version__ = "3.0.0a3"
+__version__ = "3.0.0a4"
 
 ROOT: Final = Path(__file__).parents[1]
 

poptimizer/controllers/bus/events.py

@@ -6,6 +6,7 @@
 from poptimizer.use_cases.dl.features import index as index_features
 from poptimizer.use_cases.dl.features import quotes as quotes_features
 from poptimizer.use_cases.dl.features import securities as tickers_features
+from poptimizer.use_cases.dl.features import year_day as year_day_features
 from poptimizer.use_cases.evolve import evolve
 from poptimizer.use_cases.moex import data, index, quotes, securities, usd
 from poptimizer.use_cases.portfolio import forecasts, portfolio
@@ -25,6 +26,7 @@ def register_handlers(
     bus.register_event_handler(portfolio.PortfolioHandler(), msg.IndefiniteRetryPolicy)
     bus.register_event_handler(quotes_features.QuotesFeatHandler(), msg.IndefiniteRetryPolicy)
     bus.register_event_handler(index_features.IndexesFeatHandler(), msg.IndefiniteRetryPolicy)
+    bus.register_event_handler(year_day_features.YearDayFeatHandler(), msg.IndefiniteRetryPolicy)
     bus.register_event_handler(tickers_features.SecFeatHandler(), msg.IndefiniteRetryPolicy)
     bus.register_event_handler(status.DivStatusHandler(http_client), msg.IgnoreErrorsPolicy)
     bus.register_event_handler(reestry.ReestryHandler(http_client), msg.IgnoreErrorsPolicy)

poptimizer/domain/dl/features.py

@@ -40,9 +40,27 @@ def _value_less_than_size(self) -> Self:
         return self
 
 
+@unique
+class EmbSeqFeat(StrEnum):
+    year_day = auto()
+
+
+class EmbeddingSeqFeatDesc(BaseModel):
+    sequence: list[NonNegativeInt]
+    size: int = Field(ge=2)
+
+    @model_validator(mode="after")
+    def _value_less_than_size(self) -> Self:
+        if any(value >= self.size for value in self.sequence):
+            raise ValueError("embedding value not less size")
+
+        return self
+
+
 class Features(domain.Entity):
     numerical: list[dict[NumFeat, FiniteFloat]] = Field(default_factory=list)
     embedding: dict[EmbFeat, EmbeddingFeatDesc] = Field(default_factory=dict)
+    embedding_seq: dict[EmbSeqFeat, EmbeddingSeqFeatDesc] = Field(default_factory=dict)
 
     @field_validator("numerical")
     def _numerical_match_labels(
@@ -58,11 +76,24 @@ def _numerical_match_labels(
 
         return numerical
 
+    @model_validator(mode="after")
+    def _embedding_seq_len_match_numerical(self) -> Self:
+        if not self.embedding_seq:
+            return self
+
+        num_len = len(self.numerical)
+        for desc in self.embedding_seq.values():
+            if len(desc.sequence) != num_len:
+                raise ValueError("embedding sequence length mismatch")
+
+        return self
+
     def _check_new_day(self, day: domain.Day) -> None:
         if self.day != day:
             self.day = day
             self.numerical.clear()
             self.embedding.clear()
+            self.embedding_seq.clear()
 
     def update_numerical(self, day: domain.Day, num_feat_df: pd.DataFrame) -> None:
         self._check_new_day(day)

poptimizer/use_cases/dl/features/index.py

@@ -20,7 +20,7 @@ async def __call__(self, ctx: handler.Ctx, msg: handler.QuotesFeatUpdated) -> ha
             for pos in port.positions:
                 tg.create_task(_add_indexes_features(ctx, domain.UID(pos.ticker), indexes))
 
-        return handler.IndexFeatUpdated(day=msg.day)
+        return handler.IndexFeatUpdated(trading_days=msg.trading_days)
 
 
 async def _load_indexes(ctx: handler.Ctx, df_index: pd.DatetimeIndex) -> list[dict[features.NumFeat, FiniteFloat]]:

poptimizer/use_cases/dl/features/securities.py

@@ -8,7 +8,7 @@
 
 
 class SecFeatHandler:
-    async def __call__(self, ctx: handler.Ctx, msg: handler.IndexFeatUpdated) -> handler.SecFeatUpdated:
+    async def __call__(self, ctx: handler.Ctx, msg: handler.YearDayFeatUpdated) -> handler.SecFeatUpdated:
         async with asyncio.TaskGroup() as tg:
             sec_task = tg.create_task(ctx.get(securities.Securities))
             port = await ctx.get(portfolio.Portfolio)

poptimizer/use_cases/dl/features/year_day.py

@@ -0,0 +1,26 @@
+import asyncio
+
+from poptimizer.domain import domain
+from poptimizer.domain.dl.features import EmbeddingSeqFeatDesc, EmbSeqFeat, Features
+from poptimizer.domain.portfolio import portfolio
+from poptimizer.use_cases import handler
+
+
+class YearDayFeatHandler:
+    async def __call__(self, ctx: handler.Ctx, msg: handler.IndexFeatUpdated) -> handler.YearDayFeatUpdated:
+        async with asyncio.TaskGroup() as tg:
+            port = await ctx.get(portfolio.Portfolio)
+
+            for pos in port.positions:
+                tg.create_task(_create_year_day_feat(ctx, domain.UID(pos.ticker), msg.trading_days))
+
+        return handler.YearDayFeatUpdated(day=msg.day)
+
+
+async def _create_year_day_feat(ctx: handler.Ctx, ticker: domain.UID, trading_days: domain.TradingDays) -> None:
+    feat = await ctx.get_for_update(Features, ticker)
+
+    feat.embedding_seq[EmbSeqFeat.year_day] = EmbeddingSeqFeatDesc(
+        sequence=[trading_days[-n].timetuple().tm_yday for n in reversed(range(1, len(feat.numerical) + 1))],
+        size=366,
+    )

poptimizer/use_cases/handler.py

@@ -74,6 +74,15 @@ def day(self) -> domain.Day:
 
 
 class IndexFeatUpdated(Event):
+    trading_days: domain.TradingDays = Field(repr=False)
+
+    @computed_field
+    @property
+    def day(self) -> domain.Day:
+        return self.trading_days[-1]
+
+
+class YearDayFeatUpdated(Event):
     day: domain.Day