csv_2_plot: add support for data filtering with multiple filter types and window size customization

Signed-off-by: YoungSoo Shin <shinys000114@gmail.com>

example/logger/csv_2_plot.py

@@ -7,11 +7,14 @@ import pandas as pd
 from dateutil.tz import gettz
 from matplotlib.ticker import MultipleLocator, FuncFormatter
 import math
+from scipy.signal import savgol_filter
+from scipy.ndimage import gaussian_filter1d
 
 
 def plot_power_data(csv_path, output_path, plot_types, sources,
                     voltage_y_max=None, current_y_max=None, power_y_max=None,
-                    relative_time=False, time_x_line=None, time_x_label=None):
+                    relative_time=False, time_x_line=None, time_x_label=None,
+                    filter_type=None, window_size=None):
     """
     Reads power data from a CSV file and generates a plot image.
 
@@ -26,6 +29,8 @@ def plot_power_data(csv_path, output_path, plot_types, sources,
         relative_time (bool): If True, the x-axis will show elapsed time from the start.
         time_x_line (float, optional): Interval in seconds for x-axis grid lines.
         time_x_label (float, optional): Interval in seconds for x-axis labels.
+        filter_type (str, optional): The type of filter to apply.
+        window_size (int, optional): The window size for the filter in seconds.
     """
     try:
         # Read the CSV file into a pandas DataFrame
@@ -57,10 +62,17 @@ def plot_power_data(csv_path, output_path, plot_types, sources,
         return
 
     # --- Calculate Average Interval ---
-    avg_interval_ms = 0
+    avg_interval_s = 0
     if len(df) > 1:
         avg_interval = df['timestamp'].diff().mean()
-        avg_interval_ms = avg_interval.total_seconds() * 1000
+        avg_interval_s = avg_interval.total_seconds()
+        avg_interval_ms = avg_interval_s * 1000
+
+    # --- Auto-set window size if not provided ---
+    if filter_type and window_size is None and avg_interval_s > 0:
+        window_size = avg_interval_s * 15  # Default to 15x the average interval
+        print(f"Window size not specified. Automatically set to {window_size:.2f} seconds.")
+
 
     # --- Calculate Average Voltages ---
     avg_voltages = {}
@@ -105,10 +117,46 @@ def plot_power_data(csv_path, output_path, plot_types, sources,
         max_data_value = 0
         for j, col_name in enumerate(config['cols']):
             if col_name in df.columns:
-                ax.plot(x_axis_data, df[col_name], label=channel_labels[j], color=channel_colors[j], zorder=2)
+                # Plot original data
+                ax.plot(x_axis_data, df[col_name], label=f'{channel_labels[j]} (Raw)', color=channel_colors[j], alpha=0.5, zorder=2)
                 max_col_value = df[col_name].max()
                 if max_col_value > max_data_value:
                     max_data_value = max_col_value
+
+                # --- Apply and plot filtered data ---
+                if filter_type and window_size:
+                    # Calculate window size in samples
+                    if avg_interval_s > 0:
+                        window_samples = int(window_size / avg_interval_s)
+                        if window_samples < 1:
+                            window_samples = 1
+                        
+                        filtered_data = None
+                        filter_label = ""
+
+                        if filter_type == 'savgol':
+                            # Window size for savgol must be odd
+                            if window_samples % 2 == 0:
+                                window_samples += 1
+                            if window_samples > 2: # Polyorder must be less than window_length
+                                filtered_data = savgol_filter(df[col_name], window_samples, 2) # polynomial order 2
+                                filter_label = "Savitzky-Golay"
+                        elif filter_type == 'moving_average':
+                            filtered_data = df[col_name].rolling(window=window_samples, center=True).mean()
+                            filter_label = "Moving Average"
+                        elif filter_type == 'ema':
+                            filtered_data = df[col_name].ewm(span=window_samples, adjust=False).mean()
+                            filter_label = "Exponential Moving Average"
+                        elif filter_type == 'gaussian':
+                            # Sigma is a fraction of the window size
+                            sigma = window_samples / 4.0
+                            filtered_data = gaussian_filter1d(df[col_name], sigma=sigma)
+                            filter_label = "Gaussian"
+
+
+                        if filtered_data is not None:
+                             ax.plot(x_axis_data, filtered_data, label=f'{channel_labels[j]} ({filter_label})', color=channel_colors[j], linestyle='-', linewidth=1.5, zorder=3)
+
             else:
                 print(f"Warning: Column '{col_name}' not found in CSV. Skipping.")
 
@@ -256,6 +304,16 @@ def main():
     )
     parser.add_argument("--time_x_line", type=float, help="Interval in seconds for x-axis grid lines.")
     parser.add_argument("--time_x_label", type=float, help="Interval in seconds for x-axis labels.")
+    parser.add_argument(
+        '--filter',
+        choices=['savgol', 'moving_average', 'ema', 'gaussian'],
+        help='The type of filter to apply to the data.'
+    )
+    parser.add_argument(
+        '--window_size',
+        type=float,
+        help='The window size for the filter in seconds.'
+    )
 
 
     args = parser.parse_args()
@@ -270,7 +328,9 @@ def main():
         power_y_max=args.power_y_max,
         relative_time=args.relative_time,
         time_x_line=args.time_x_line,
-        time_x_label=args.time_x_label
+        time_x_label=args.time_x_label,
+        filter_type=args.filter,
+        window_size=args.window_size
     )