From e148d01c8fe21a3d6eec01cfffb602b664d997e0 Mon Sep 17 00:00:00 2001
From: Le Roux Erwan <erwan.le-roux@irstea.fr>
Date: Mon, 5 Nov 2018 15:03:55 +0100
Subject: [PATCH] add normalized coordinates
---
.../abstract_coordinates.py | 26 ++++----
.../normalized_coordinates.py | 66 +++++++++++++++++++
2 files changed, 81 insertions(+), 11 deletions(-)
diff --git a/spatio_temporal_dataset/spatial_coordinates/abstract_coordinates.py b/spatio_temporal_dataset/spatial_coordinates/abstract_coordinates.py
index 31bf1787..ed6bca63 100644
--- a/spatio_temporal_dataset/spatial_coordinates/abstract_coordinates.py
+++ b/spatio_temporal_dataset/spatial_coordinates/abstract_coordinates.py
@@ -1,4 +1,5 @@
import os.path as op
+import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
@@ -31,31 +32,34 @@ class AbstractSpatialCoordinates(object):
df = pd.read_csv(csv_path)
return cls.from_df(df)
- def df_coord_split(self, split_str):
+ def coord_x_y_values(self, df_coord: pd.DataFrame) -> np.ndarray:
+ return df_coord.loc[:, [self.COORD_X, self.COORD_Y]].values
+
+ def df_coord_split(self, split_str: str) -> pd.DataFrame:
assert self.s_split is not None
ind = self.s_split == split_str
return self.df_coord.loc[ind]
@property
- def df_coord_train(self):
- return self.df_coord_split(self.TRAIN_SPLIT_STR)
-
- @property
- def df_coord_test(self):
- return self.df_coord_split(self.TEST_SPLIT_STR)
+ def coord(self) -> np.ndarray:
+ return self.coord_x_y_values(df_coord=self.df_coord)
@property
- def nb_points(self):
- return len(self.df_coord)
+ def coord_train(self) -> np.ndarray:
+ return self.coord_x_y_values(df_coord=self.df_coord_split(self.TRAIN_SPLIT_STR))
@property
- def coord(self):
- return self.df_coord.values
+ def coord_test(self) -> np.ndarray:
+ return self.coord_x_y_values(df_coord=self.df_coord_split(self.TEST_SPLIT_STR))
@property
def index(self):
return self.df_coord.index
+ @property
+ def nb_points(self):
+ return len(self.df_coord)
+
def visualization(self):
x, y = self.coord[:, 0], self.coord[:, 1]
plt.scatter(x, y)
diff --git a/spatio_temporal_dataset/spatial_coordinates/normalized_coordinates.py b/spatio_temporal_dataset/spatial_coordinates/normalized_coordinates.py
index e69de29b..1ef1de44 100644
--- a/spatio_temporal_dataset/spatial_coordinates/normalized_coordinates.py
+++ b/spatio_temporal_dataset/spatial_coordinates/normalized_coordinates.py
@@ -0,0 +1,66 @@
+import pandas as pd
+
+from spatio_temporal_dataset.spatial_coordinates.abstract_coordinates import AbstractSpatialCoordinates
+from spatio_temporal_dataset.spatial_coordinates.alps_station_coordinates import AlpsStationCoordinate
+
+
+class AbstractNormalizingFunction(object):
+
+ def normalize(self, df_coord: pd.DataFrame) -> pd.DataFrame:
+ assert len(df_coord.columns) == 2
+ return df_coord
+
+
+class NormalizedCoordinates(AbstractSpatialCoordinates):
+
+ @classmethod
+ def from_coordinates(cls, spatial_coordinates: AbstractSpatialCoordinates,
+ normalizing_function: AbstractNormalizingFunction):
+ df_coord_normalized = spatial_coordinates.df_coord.copy()
+ coord_XY = [spatial_coordinates.COORD_X, spatial_coordinates.COORD_Y]
+ df_coord_normalized.loc[:, coord_XY] = normalizing_function.normalize(df_coord_normalized.loc[:, coord_XY])
+ return cls(df_coord=df_coord_normalized, s_split=spatial_coordinates.s_split)
+
+
+"""
+Define various types of normalizing functions
+"""
+
+
+class UniformNormalization(AbstractNormalizingFunction):
+ """Normalize similarly the X and Y axis with a single function so as to conserve proportional distances"""
+
+ def normalize(self, df_coord: pd.DataFrame) -> pd.DataFrame:
+ df_coord = super().normalize(df_coord)
+ for i in range(2):
+ df_coord.iloc[:, i] = self.uniform_normalization(df_coord.iloc[:, i])
+ return df_coord
+
+ def uniform_normalization(self, s_coord: pd.Series) -> pd.Series:
+ return s_coord
+
+
+class BetweenZeroAndOneNormalization(UniformNormalization):
+ """Normalize such that min(coord) >= (0,0) and max(coord) <= (1,1)"""
+
+ def __init__(self) -> None:
+ self.min_coord = None
+ self.max_coord = None
+
+ def normalize(self, df_coord: pd.DataFrame) -> pd.DataFrame:
+ # Compute the min and max globally
+ self.min_coord, self.max_coord = df_coord.min().min(), df_coord.max().max()
+ # Then, call the super method that will call the uniform_normalization method
+ return super().normalize(df_coord)
+
+ def uniform_normalization(self, s_coord: pd.Series) -> pd.Series:
+ s_coord_shifted = s_coord - self.min_coord
+ s_coord_scaled = s_coord_shifted / (self.max_coord - self.min_coord)
+ return s_coord_scaled
+
+
+if __name__ == '__main__':
+ coord = AlpsStationCoordinate.from_csv()
+ normalized_coord = NormalizedCoordinates.from_coordinates(spatial_coordinates=coord,
+ normalizing_function=BetweenZeroAndOneNormalization())
+ normalized_coord.visualization()
--
GitLab