feat: output threshold + final scores

Signed-off-by: Avik Basu <[email protected]>

numalogic/backtest/_prom.py

@@ -188,7 +188,8 @@ def generate_scores(
  x_scaled = preproc_udf.compute(model=artifacts["preproc_clf"], input_=x_test)
 
  ds = StreamingDataset(x_scaled, seq_len=self.conf.window_size)
- anomaly_scores = np.zeros((len(ds), self.conf.window_size), dtype=np.float32)
+ raw_scores = np.zeros((len(ds), self.conf.window_size), dtype=np.float32)
+ final_scores = np.zeros_like(raw_scores, dtype=np.float32)
  postproc_func = PostprocessFactory().get_instance(self.conf.numalogic_conf.postprocess)
 
  x_recon = np.zeros(
@@ -198,22 +199,24 @@ def generate_scores(
  # Model Inference
  for idx, arr in enumerate(ds):
  x_recon[idx] = nn_udf.compute(model=artifacts["model"], input_=arr)
- anomaly_scores[idx], _ = postproc_udf.compute(
+ raw_scores[idx], final_scores[idx] = postproc_udf.compute(
  model=artifacts["threshold_clf"],
  input_=x_recon[idx],
  postproc_clf=postproc_func,
  )
 
  x_recon = inverse_window(torch.from_numpy(x_recon)).numpy()
- anomaly_scores = inverse_window(
- torch.unsqueeze(torch.from_numpy(anomaly_scores), dim=2)
+ raw_scores = inverse_window(torch.unsqueeze(torch.from_numpy(raw_scores), dim=2)).numpy()
+ final_scores = inverse_window(
+ torch.unsqueeze(torch.from_numpy(final_scores), dim=2)
  ).numpy()
 
  return self._construct_output(
  df_test,
  preproc_out=x_scaled,
  nn_out=x_recon,
- postproc_out=anomaly_scores,
+ thresh_out=raw_scores,
+ postproc_out=final_scores,
  )
 
  @classmethod
@@ -276,6 +279,7 @@ def _construct_output(
  input_df: pd.DataFrame,
  preproc_out: NDArray[float],
  nn_out: NDArray[float],
+ thresh_out: NDArray[float],
  postproc_out: NDArray[float],
  ) -> pd.DataFrame:
  ts_idx = input_df.index
@@ -291,6 +295,11 @@ def _construct_output(
  columns=self.metrics,
  index=ts_idx,
  ),
+ "thresh_out": pd.DataFrame(
+ thresh_out,
+ columns=["unified_score"],
+ index=ts_idx,
+ ),
  "postproc_out": pd.DataFrame(
  postproc_out,
  columns=["unified_score"],