cast as anypath on load and nowhere else

cyano/cli.py

@@ -1,5 +1,6 @@
 import yaml
 
+from cloudpathlib import AnyPath
 from pathlib import Path
 import typer
 
@@ -35,6 +36,8 @@ def predict(
     """Load an existing cyanobacteria prediction model and generate
     severity level predictions for a set of samples.
     """
+    samples_path = AnyPath(samples_path)
+
     pipeline = CyanoModelPipeline.from_disk(model_zip)
     pipeline.run_prediction(samples_path, output_path)
 
@@ -53,6 +56,9 @@ def evaluate(
         default=Path.cwd() / "metrics", help="Folder in which to save out metrics and plots."
     ),
 ):
+    y_pred_csv = AnyPath(y_pred_csv)
+    y_true_csv = AnyPath(y_true_csv)
+
     EvaluatePreds(
         y_pred_csv=y_pred_csv, y_true_csv=y_true_csv, save_dir=save_dir
     ).calculate_all_and_save()

cyano/evaluate.py

@@ -1,7 +1,6 @@
 import json
 from pathlib import Path
 
-from cloudpathlib import AnyPath
 import pandas as pd
 import lightgbm as lgb
 import matplotlib.pyplot as plt
@@ -88,7 +87,7 @@ def __init__(self, y_true_csv: Path, y_pred_csv: Path, save_dir: Path, model: lg
         """
         self.model = model
 
-        y_true_df = pd.read_csv(AnyPath(y_true_csv))
+        y_true_df = pd.read_csv(y_true_csv)
 
         if "severity" not in y_true_df.columns:
             raise ValueError("Evaluation data must include a `severity` column to evaluate.")
@@ -98,7 +97,7 @@ def __init__(self, y_true_csv: Path, y_pred_csv: Path, save_dir: Path, model: lg
         self.y_true = y_true_df["severity"].rename("y_true")
         self.metadata = y_true_df.drop(columns=["severity"])
 
-        y_pred_df = pd.read_csv(AnyPath(y_pred_csv)).set_index("sample_id")
+        y_pred_df = pd.read_csv(y_pred_csv).set_index("sample_id")
 
         try:
             self.y_pred = y_pred_df.loc[self.y_true.index]["severity"].rename("y_pred")

cyano/experiment/experiment.py

@@ -2,8 +2,9 @@
 from typing import Union
 import yaml
 
+from cloudpathlib import AnyPath
 from loguru import logger
-from pydantic import BaseModel, ConfigDict, field_serializer
+from pydantic import BaseModel, ConfigDict, field_serializer, field_validator
 
 from cyano.config import FeaturesConfig, ModelTrainingConfig
 from cyano.pipeline import CyanoModelPipeline
@@ -20,6 +21,10 @@ class ExperimentConfig(BaseModel):
     debug: bool = False
     num_processes: int = 4
 
+    @field_validator("train_csv", "predict_csv")
+    def convert_filepaths(cls, path_field):
+        return AnyPath(path_field)
+
     # Avoid conflict with pydantic protected namespace
     model_config = ConfigDict(protected_namespaces=())
 

cyano/pipeline.py

@@ -4,7 +4,6 @@
 import yaml
 from zipfile import ZipFile
 
-from cloudpathlib import AnyPath
 import lightgbm as lgb
 from loguru import logger
 import pandas as pd
@@ -41,7 +40,7 @@ def __init__(
 
     def _prep_train_data(self, data, debug: bool = False):
         """Load labels and save out samples with UIDs"""
-        labels = pd.read_csv(AnyPath(data))
+        labels = pd.read_csv(data)
         labels = labels[["date", "latitude", "longitude", "severity"]]
         labels = add_unique_identifier(labels)
         if debug:
@@ -79,8 +78,9 @@ def _prepare_features(self, samples):
         features = generate_features(samples, satellite_meta, self.features_config, self.cache_dir)
         save_features_to = self.cache_dir / "features_train.csv"
         features.to_csv(save_features_to, index=True)
+        pct_with_features = features.index.nunique() / samples.shape[0]
         logger.success(
-            f"{features.shape[1]:,} features for {features.index.nunique():,} samples (of {samples.shape[0]:,}) saved to {save_features_to}"
+            f"{features.shape[1]:,} features for {features.index.nunique():,} samples ({pct_with_features:.0%}) saved to {save_features_to}"
         )
 
         return features
@@ -123,7 +123,7 @@ def from_disk(cls, filepath, cache_dir=None):
         return cls(features_config=features_config, model=model, cache_dir=cache_dir)
 
     def _prep_predict_data(self, data, debug: bool = False):
-        df = pd.read_csv(AnyPath(data))
+        df = pd.read_csv(data)
         df = add_unique_identifier(df)
 
         samples = df[["date", "latitude", "longitude"]]

tests/assets/evaluate_data.csv

@@ -1,4 +1,4 @@
-latitude,longitude,date,split,region,severity,density
+latitude,longitude,date,split,region,severity,density_cells_per_ml
 40.090275,-76.873132,2018-05-21,train,northeast,1,0.0
 35.7200811863161,-79.1374207771809,2013-05-22,train,south,2,29046.0
 35.6940254103693,-79.1858165585188,2016-10-18,train,south,1,94.0