added interpolate_signal_loc() in rtr process

measprocess/rtr/process.py

 import os.path as path
 import pandas as pd
 import numpy as np
+import geopandas as gpd
+from scipy import interpolate
 
-def process_4G(details, path_to_store_files):
+def process(details, path_to_store_files):
+    '''
+    Generates 2 csv files: signal.csv and location.csv and stores them in a folder defined under path_to_store_files
+
+    :param details: details returned by the function rtr_details() from fetch.py in rtr
+    :param path_to_store_files: existing path at which the generated csv files will be saved to
+    '''
     d = pd.DataFrame(details)
     path_processed_signal   = path.join(path_to_store_files,'signal.csv')
     path_processed_location   = path.join(path_to_store_files,'location.csv')
@@ -52,4 +60,72 @@ def process_4G(details, path_to_store_files):
         tests_locations, ignore_index=True
     ).to_csv(path_processed_location, index=False)
 
-    print('\t\tDone.')
\ No newline at end of file
+    print('\t\tDone.')
+
+def interpolate_signal_loc(path_processed_signal, path_processed_location, path_processed_final=None):
+    '''
+    Interpolates gps location values based on path_processed_location for each not NaN lte_rsrp value.
+
+    :param path_processed_signal: path of signal.csv
+    :param path_processed_location: path of location.csv
+    :path_processed_final: path of final.csv where the returned dataframe is stored, this is optional, default is None
+
+    :return df_signal_final: pandas dataframe containing all not NaN values from signal.csv, 
+                             including interpolated longitude and latitude for each value
+    '''
+
+    df_signal = pd.read_csv(path_processed_signal)
+    df_loc = pd.read_csv(path_processed_location)
+    
+    df_signal = df_signal[df_signal['cat_technology']=='4G'] #select only 4G techonology
+    df_signal = df_signal[df_signal['lte_rsrp'].notna()] # remove all where lte_rsrp value is missing
+
+    #grouping both dfs by id
+    df_signal_grouped = df_signal.groupby(['id'])
+    df_loc_grouped = df_loc.groupby(['id'])
+
+    #loop over each uuid in df_signal_grouped
+    df_signal_final=[]
+    i = 0
+    for ID, df_signal_selected in df_signal_grouped:
+        i=i+1
+        if (i % 50) == 0:
+            print('\t\t{}/{}'.format(i, len(df_signal_grouped)))
+
+        df_loc_selected = df_loc_grouped.get_group(ID) #get locations with same uuid
+
+        if len(df_loc_selected)<2: #we need at least 2 locations for interpolation, otherwise skip it
+            continue
+
+        lon_lat = gpd.GeoDataFrame(geometry=gpd.points_from_xy(df_loc_selected['long'], df_loc_selected['lat']))
+        timeseries_loc = df_loc_selected['time_elapsed'].to_numpy()
+        timeseries_signal = df_signal_selected['time_elapsed'].to_numpy()
+
+        # delete those where we would have to extrapolate
+        S = [i for i in timeseries_signal if i >= timeseries_loc.min() and i <= timeseries_loc.max()]
+        df_signal_selected = df_signal_selected[df_signal_selected['time_elapsed'].isin(S)]
+        timeseries_signal = df_signal_selected['time_elapsed'].to_numpy()
+
+        # do interpolation in EPSG:31287
+        lon_lat = lon_lat.set_crs("EPSG:4326")
+        x_y = lon_lat.to_crs("EPSG:31287")
+        interp_x = interpolate.interp1d(timeseries_loc, x_y.geometry.x.to_numpy())
+        xnew = interp_x(timeseries_signal)
+        interp_y = interpolate.interp1d(timeseries_loc, x_y.geometry.y.to_numpy())
+        ynew = interp_y(timeseries_signal)
+        df_signal_selected['x'] = xnew
+        df_signal_selected['y'] = ynew
+
+        df_signal_final.append(df_signal_selected)
+    
+    df_signal_final = pd.concat(df_signal_final, ignore_index=True)
+    xy = gpd.GeoDataFrame(geometry=gpd.points_from_xy(df_signal_final.x, df_signal_final.y)).set_crs('EPSG:31287').to_crs('EPSG:4326')
+    df_signal_final.drop(['x','y'], axis=1, inplace=True) #remove x and y in epsg:31287
+    # add long and lat in epsg:4326
+    df_signal_final['long'] = xy.geometry.x
+    df_signal_final['lat'] = xy.geometry.y
+
+    if path_processed_final!=None:
+        df_signal_final.to_csv(path_processed_final, index=False)
+
+    return df_signal_final
\ No newline at end of file