initial files

Author: X-Bowen

Instance_Creation_sequence1.py

@@ -0,0 +1,149 @@
+import numpy as np
+import pickle
+from datetime import datetime
+import os
+
+# Change the current working directory to the location of 'Combined Trajectory_Label_Geolife' folder.
+os.chdir(r'/home/ubuntu/Transport-Mode-GPS-CNN/data/Combined_Trajectory_Label_Geolife')
+
+# create 'daysDate' function to convert start and end time to a number of days
+
+
+def days_date(time_str):
+    date_format = "%Y/%m/%d %H:%M:%S"
+    current = datetime.strptime(time_str, date_format)
+    date_format = "%Y/%m/%d"
+    bench = datetime.strptime('1899/12/30', date_format)
+    no_days = current - bench
+    delta_time_days = no_days.days + current.hour / 24.0 + current.minute / (24. * 60.) + current.second / (24. * 3600.)
+    return delta_time_days
+
+# Change Mode Name to Mode index
+Mode_Index = {"walk": 0, "run": 9, "bike": 1, "bus": 2, "car": 3, "taxi": 3, "subway": 4, "railway": 4,
+              "train": 4, "motocycle": 8, "boat": 9, "airplane": 9, "other": 9}
+
+## 0: walk 
+## 1: bike 
+## 2: bus
+## 3: car, taxi
+## 4: subway, railway, train 
+## 8: motocycle 
+## 9: run, boat, airplane, other
+
+# Ground modes are the modes that we use in this paper.
+Ground_Mode = ['walk', 'bike', 'bus', 'car', 'taxi', 'subway', 'railway', 'train']
+## Ground_Mode = [0,     1,      2,      3,      3,      4,          4,         4]
+
+# Trajectory_Array and Label_Array are the final lists which each of its element is for one user
+Trajectory_Array = []
+Label_Array = []
+Trajectory_Label_Array = []
+UserNon = range(190)  ## 0-179, might not be enough
+
+# 1
+for k in UserNon: 
+## will be the same
+#for k in range(len(UserNon)):
+    
+    InputFile = "combined" + str(k) + ".plt"
+    ##InputFile = "combined" + str(UserNon[k]) + ".plt"
+    table = []
+    try:
+        ## with open(InputFile, 'r') as inp:
+        with open(InputFile, 'rb') as inp:
+            for row in inp:
+                row = row.rstrip()
+                row = row.decode("utf-8")
+                row = row.split(',')
+                if len(row) == 7:
+                    table.append(row)
+    except IOError:
+        continue
+    #print(table)
+
+# TrajectoryMatrix = contains lat, long, date in each column
+    table_array = np.array(table, dtype=object)
+    #print (table_array.shape)
+    # for 2d array, axis = -1 is equal to axis = 1
+    TrajectoryMatrix = np.stack((table_array[:, 0], table_array[:, 1], table_array[:, 4]), axis=-1)
+    #print( TrajectoryMatrix.shape)
+    for i in range(len(table_array[:, 0])):
+        for j in range(3):
+            TrajectoryMatrix[i, j] = float(TrajectoryMatrix[i, j])
+
+    Trajectory_Array.append(TrajectoryMatrix)
+
+
+# end1
+# 2.Modify the labels file and create array with start_time, end_time in days and labels
+    InputFile = "labels" + str(k) + ".txt"
+    table = []
+    with open(InputFile, 'rb') as inp:
+        for row in inp:
+            row = row.rstrip()
+            row = row.decode("utf-8")
+            row = row.split('\t')
+            if len(row) == 3:
+                table.append(row)
+
+    LabelFile = np.array(table, dtype=object)
+
+
+# StartTime and EndTime in days after 1899/12/30 for each data point in labels.cv
+# Modify label for those rows that don't have any time and labels
+# LabelMatrix = the array that has Start time(days), End time(days), and labels
+
+    StartTime = []
+    EndTime = []
+    label = []
+    Error = []
+
+    for i in range(len(LabelFile[:, 0])):
+        try:
+            if LabelFile[i, 2] in Ground_Mode:
+                StartTime.append(days_date(LabelFile[i, 0]))
+                EndTime.append(days_date(LabelFile[i, 1]))
+                label.append(Mode_Index[LabelFile[i, 2]])
+        except ValueError:
+            Error.append(i)
+
+    LabelMatrix = (np.vstack((StartTime, EndTime, label))).T
+    Label_Array.append(LabelMatrix)
+
+    # End2
+
+
+
+    # 3.Assign the labels to the trajectories
+    # Trajectory = zip(lat, long, date)
+    Dates = np.split(TrajectoryMatrix, 3, axis=-1)[2]
+    Sec = 1 / (24.0 * 3600.0)
+    # C_list: all the rows in the TrajectoryMatrix that should be picked up
+    C_list = []
+    # Mode_Trajectory: all labels
+    Mode_Trajectory = []
+    for index, row in enumerate(LabelMatrix):
+        A = np.where(Dates >= (float(row[0]) - Sec))
+        B = np.where(Dates <= (float(row[1]) + Sec))
+        C = list(set(A[0]).intersection(B[0]))
+        if len(C) == 0:
+            print("error")
+        [Mode_Trajectory.append(row[2]) for i in C]
+        [C_list.append(i) for i in C]
+
+    TrajectoryMatrix = [TrajectoryMatrix[i, :] for i in C_list]
+    TrajectoryMatrix = np.array(TrajectoryMatrix)
+    Mode_Trajectory = np.array(Mode_Trajectory)
+    Trajectory_Label = (np.vstack((TrajectoryMatrix.T, Mode_Trajectory))).T
+
+    Trajectory_Label_Array.append(Trajectory_Label)
+
+    # End3
+
+##for i in Trajectory_Label_Array:
+##    Shape = np.shape(i)
+##    print(Shape, "+++")
+
+# Save Trajectory_Array and Label_Array for all users
+with open("/home/ubuntu/Transport-Mode-GPS-CNN/data/1_Trajectory_Label_Array.pickle", 'wb') as f:  # Python 3: open(..., 'wb')
+    pickle.dump(Trajectory_Label_Array, f)