Encoding (#33)

.pre-commit-config.yaml

@@ -12,7 +12,6 @@ repos:
       - id: trailing-whitespace
       - id: end-of-file-fixer
       - id: check-yaml
-      - id: check-added-large-files
   - repo: https://github.com/PyCQA/isort
     rev: 5.10.1
     hooks:

MANIFEST.in

@@ -1 +1,2 @@
 include src/rsdiv/embedding/*.pkl
+include src/rsdiv/encoding/*.json

README.md

@@ -104,6 +104,17 @@ metrics.get_lorenz_curve(ratings['movieId'])
 ```
 ![Lorenz](pics/Lorenz.png)
 
+### Evaluate the unblance from a sense of location
+**rsdiv** provides the encoders including geography encoding function to improve the intuitive understanding for practitioners, to start with the random values:
+```
+>>> geo = rs.GeoEncoder()
+>>> df = geo.read_source()
+>>> rng = np.random.RandomState(42)
+>>> df['random_values'] = rng.rand(len(df))
+>>> geo.graw_geo_graph(df, 'random_values')
+```
+![GeoEncoder](pics/random_values.png)
+
 ### Train a recommender
 **rsdiv** provides various implementations of core recommender algorithms. To start with, a wrapper for `LightFM` is also supported:
 ```
@@ -139,6 +150,7 @@ Not only for categorical labels, **rsdiv** also supports embedding for items, fo
 - implement the Bounded Greedy Selection Strategy, BGS diversify algorithm
 - implement the Determinantal Point Process, DPP diversify algorithm
 - implement the Modified Gram-Schmidt, MGS diversify algorithm
+
 ### Hyperparameter optimization
 **TODO**
 - compatible with [Optuna](https://github.com/optuna/optuna).

setup.py

@@ -15,12 +15,13 @@
         "lightfm>=1.16",
         "scikit-learn>=1.1.1",
         "matplotlib>=3.5.2",
+        "plotly>=5.6.0",
     ],
 )
 
 setup(
     name="rsdiv",
-    version="0.1.9",
+    version="0.1.10",
     author="Yin Cheng",
     author_email="yin.sjtu@gmail.com",
     long_description=LONG_DESCRIPTION,

src/rsdiv/__init__.py

@@ -1,4 +1,5 @@
 from .dataset import *
 from .embedding import *
+from .encoding import *
 from .evaluation import *
 from .recommenders import *

src/rsdiv/embedding/base.py

@@ -1,5 +1,4 @@
 from abc import ABCMeta
-from pathlib import Path
 from typing import Dict
 
 import numpy as np

src/rsdiv/encoding/__init__.py

@@ -0,0 +1,9 @@
+import imp
+
+from .base import BaseEncoder
+from .geo_encoder import GeoEncoder
+
+__all__ = [
+    "BaseEncoder",
+    "GeoEncoder",
+]

src/rsdiv/encoding/base.py

@@ -0,0 +1,10 @@
+from abc import ABCMeta, abstractclassmethod, abstractmethod
+from typing import Any, Dict, List, Union
+
+
+class BaseEncoder(metaclass=ABCMeta):
+    encode_source: Dict[str, Any]
+
+    @abstractmethod
+    def encoding_single(cls, org: Union[List, str]) -> Union[int, str]:
+        raise NotImplementedError("embedding_single must be implemented.")

src/rsdiv/encoding/geo_encoder.py

@@ -0,0 +1,65 @@
+import json
+import pkgutil
+from typing import Any, Dict, List, Optional, Union
+
+import numpy as np
+import pandas as pd
+import plotly.express as px
+from scipy import spatial
+
+from .base import BaseEncoder
+
+
+class GeoEncoder(BaseEncoder):
+    r"""Plotly Sample Datasets."""
+    ECD_PATH: Optional[bytes] = pkgutil.get_data(
+        "rsdiv.encoding", "geojson-counties-fips.json"
+    )
+    if ECD_PATH:
+        encode_source: Dict[str, Any] = json.loads(ECD_PATH)
+
+    def __init__(self) -> None:
+        super().__init__()
+        self.encoder: pd.DataFrame = self.read_source()
+        self.coord: List[np.ndarray] = self.encoder.coord.to_list()
+        self.index: pd.Index = pd.Index(self.encoder["index"])
+
+    def read_source(self) -> pd.DataFrame:
+        geo_county_dict: Dict[str, List] = {}
+        for item in self.encode_source["features"]:
+            coordinates = item["geometry"]["coordinates"]
+            parts = []
+            for part in coordinates:
+                parts.append(np.asarray(part).squeeze().mean(axis=0).squeeze())
+            coord = np.asarray(parts).mean(axis=0)[::-1]  # reverse lat/lng
+            name = item["properties"]["NAME"]
+            lsad = item["properties"]["LSAD"]
+            id = item["id"]
+            geo_county_dict[id] = [coord, name, lsad]
+        dataframe = pd.DataFrame.from_dict(
+            geo_county_dict, orient="index", columns=["coord", "name", "lstd"]
+        ).reset_index()
+        return dataframe
+
+    def encoding_single(self, org: Union[List, str]) -> Union[int, str]:
+        tree = spatial.KDTree(self.coord)
+        return str(self.index[int(tree.query(org)[1])])
+
+    def encoding_series(self, series: pd.Series) -> pd.Series:
+        encodings = pd.Series(series.apply(lambda x: self.encoding_single(x)))
+        return encodings
+
+    def graw_geo_graph(self, dataframe: pd.DataFrame, source_name: str) -> None:
+        max_value: float = np.ceil(dataframe[source_name].max())
+        min_value: float = np.floor(dataframe[source_name].min())
+        fig = px.choropleth(
+            dataframe,
+            geojson=self.encode_source,
+            locations="index",
+            color=source_name,
+            color_continuous_scale="OrRd",
+            range_color=(min_value, max_value),
+            scope="usa",
+        )
+        fig.update_layout(margin={"r": 0, "t": 0, "l": 0, "b": 0})
+        fig.show()