From fad991a00711be3176323bce5c7f0d2c3159b772 Mon Sep 17 00:00:00 2001
From: Calvin Che <cheziyi@gmail.com>
Date: Sat, 31 Aug 2024 02:28:51 +0800
Subject: [PATCH] Update to LinearScaler

---
 .../scaler/PredefinedMinMaxScaler             | 29 ----------
 .../data_processing/scaler/linear_scaler.py   | 56 +++++++++++++++++++
 2 files changed, 56 insertions(+), 29 deletions(-)
 delete mode 100644 src/synnax_shared/data_processing/scaler/PredefinedMinMaxScaler
 create mode 100644 src/synnax_shared/data_processing/scaler/linear_scaler.py

diff --git a/src/synnax_shared/data_processing/scaler/PredefinedMinMaxScaler b/src/synnax_shared/data_processing/scaler/PredefinedMinMaxScaler
deleted file mode 100644
index e9cdc92..0000000
--- a/src/synnax_shared/data_processing/scaler/PredefinedMinMaxScaler
+++ /dev/null
@@ -1,29 +0,0 @@
-import pandas as pd
-
-class PredefinedMinMaxScaler:
-    '''Scaler that scales data to the user defined range [min_val, max_val]'''
-    def __init__(self, min_val: float, max_val: float) -> None:
-        self.min_val = min_val
-        self.max_val = max_val
-
-    def scale_column(
-        self, column: pd.Series, min_val: float, max_val: float
-    ) -> pd.Series:
-        return (column - min_val) / (max_val - min_val)
-
-    def unscale_column(
-        self, column: pd.Series, min_val: float, max_val: float
-    ) -> pd.Series:
-        return column * (max_val - min_val) + min_val
-
-    def transform(self, data: pd.DataFrame) -> pd.DataFrame:
-        data = data.copy()
-        for column in data.select_dtypes(include=[np.number]):
-            data[column] = self.scale_column(data[column], self.min_val, self.max_val)
-        return data
-
-    def inverse_transform(self, data: pd.DataFrame) -> pd.DataFrame:
-        data = data.copy()
-        for column in data.select_dtypes(include=[np.number]):
-            data[column] = self.unscale_column(data[column], self.min_val, self.max_val)
-        return data
\ No newline at end of file
diff --git a/src/synnax_shared/data_processing/scaler/linear_scaler.py b/src/synnax_shared/data_processing/scaler/linear_scaler.py
new file mode 100644
index 0000000..73f0924
--- /dev/null
+++ b/src/synnax_shared/data_processing/scaler/linear_scaler.py
@@ -0,0 +1,56 @@
+from typing import TypedDict
+from numpy import number
+from pandas import DataFrame, Series
+
+
+class LinearScalerDto(TypedDict):
+    offset: float
+    divisor: float
+
+
+class LinearScaler:
+
+    def __init__(self, offset: float, divisor: float) -> None:
+        """
+        Initializes the LinearScaler.
+        E.g. To scale data from -100 to 100 to 0 to 1, the offset would be 100.0 and the divisor would be 200.0.
+
+        Parameters:
+            offset (float):
+                The offset to be applied to the data
+            divisor (float):
+                The divisor to be applied to the data
+        """
+        if divisor == 0:
+            raise ValueError("Divisor cannot be 0")
+        self.offset = offset
+        self.divisor = divisor
+
+    def transform_series(self, series: Series) -> Series:
+        float_series = series.astype(float)
+        return (float_series + self.offset) / self.divisor
+
+    def inverse_transform_series(self, series: Series) -> Series:
+        float_series = series.astype(float)
+        return (float_series * self.divisor) - self.offset
+
+    def transform_dataframe(self, dataframe: DataFrame) -> DataFrame:
+        dataframe_copy = dataframe.copy()
+        for column in dataframe_copy.select_dtypes(include=[number]):
+            dataframe_copy[column] = self.transform_series(dataframe_copy[column])
+        return dataframe_copy
+
+    def inverse_transform_dataframe(self, dataframe: DataFrame) -> DataFrame:
+        dataframe_copy = dataframe.copy()
+        for column in dataframe_copy.select_dtypes(include=[number]):
+            dataframe_copy[column] = self.inverse_transform_series(
+                dataframe_copy[column]
+            )
+        return dataframe_copy
+
+    def toDto(self) -> LinearScalerDto:
+        return {"offset": self.offset, "divisor": self.divisor}
+
+    @staticmethod
+    def fromDto(dto: LinearScalerDto):
+        return LinearScaler(dto["offset"], dto["divisor"])