finish mov_avg

filter_dev/filter/base.py

@@ -33,7 +33,7 @@ class Filter:
         calib_factor = 100. / float(self.calib_entry.get())
         df = self.device.data
         
-        df = df[df['weights'] < 10e9]
+        df = df[df['weights'] < 10e8]
         df['timestamps'] -= df['timestamps'].min()
         
         df['filtered'], df['filtered_calib'] = self.filter(df, calib_factor)

filter_dev/filter/mov_avg.py

@@ -1,21 +1,58 @@
-from tkinter import ttk
+from statistics import mean
 
 import pandas as pd
 
 from .base import Filter
 from ..gui import Slider
+from ..config import MOV_AVG_DEFAULTS
 
 class MovAvg(Filter):
     
     def init_params(self, toolbar):
         self.param_map = {
-            "window_size": Slider(toolbar, "Window Size", 1, 100, 10, self.callback),
-            "decimals": Slider(toolbar, "Decimals", 1, 5, 1, self.callback),
-            # "reset_threshold": Slider(self.toolbar, "Reset Threshold", 0.001, 0.1, 0.1, self.update),
+            "window_size": Slider(toolbar, "Window Size", 1, 100,
+                                  MOV_AVG_DEFAULTS['window_size'],
+                                  self.callback),
+            "decimals": Slider(toolbar, "Decimals", 1, 5,
+                               MOV_AVG_DEFAULTS['decimals'],
+                               self.callback),
+            "reset_threshold": Slider(toolbar, "Reset Threshold", 0.01, 1,
+                                      MOV_AVG_DEFAULTS['reset_threshold'],
+                                      self.callback, float),
+            "ignore_samples": Slider(toolbar, "Ignore Samples before reset", 1, 10,
+                                     MOV_AVG_DEFAULTS['ignore_samples'],
+                                     self.callback)
         }
     
     def filter(self, df: pd.DataFrame, calib_factor: float) -> pd.Series:
         params = self._get_params()
-        mov_avg = df['weights'].rolling(window=int(params['window_size'])).mean()
+        reset_threshold = params['reset_threshold'] / calib_factor
+
+        window = []
+        mov_avg = []
+        ignored_samples = 0
+
+        for w in df['weights']:
+            if len(window) < params['window_size']:
+                window.append(w)
+                mov_avg.append(mean(window))
+            else:
+                out_of_threshold = abs(mov_avg[-1] - w) > reset_threshold
+                if out_of_threshold and\
+                   ignored_samples < params['ignore_samples']:
+                   ignored_samples += 1
+                   mov_avg.append(mov_avg[-1])
+
+                elif out_of_threshold:
+                    ignored_samples = 0
+                    window = [w]
+                    mov_avg.append(w)
+                
+                else:
+                    ignored_samples = 0
+                    window.append(w)
+                    mov_avg.append(mean(window))
+
+        mov_avg = pd.Series(mov_avg)
         mov_avg_calib = (mov_avg * calib_factor).round(int(params['decimals']))
         return mov_avg, mov_avg_calib