#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sat Nov 18 13:02:24 2017
@author: rv
"""
import numpy as np
from matplotlib import pyplot as plt
from exploration import loadSubject
data = loadSubject(101, interpolate=False)
desc = data.describe().T
desc = desc.round(2)
print(desc)
act_desc = []
for act in data['activity'].unique():
data_act = data.loc[data['activity'] == act]
desc_ = data_act.describe().T
act_desc.append((act, desc_))
data_ = data.drop(['time', 'activity'], axis=1)
features = data_.columns.unique()
for feature in features:
m = np.zeros(len(act_desc))
s = np.zeros(len(act_desc))
import pandas as pd
import matplotlib.pyplot as plt
import numpy as np
from sklearn.ensemble import RandomForestClassifier
from sklearn.model_selection import cross_val_score
from exploration import loadSubject, slidingWindows
from summarizingData import summarizeWindows
"""----------------------------------------------------------------------------
-------------------------------------------------------------------------------
----------------------------------------------------------------------------"""
nEstimators_ = [5, 9, 15, 25]
data = loadSubject(101)
""" Choose which features to use. Outcomment all for all """
#data = data[['time','activity','hand_acc_16g_x','hand_acc_16g_y','hand_acc_16g_z']] #TRY ONLY WITH ACC HAND
#data = data[['time','activity','heart_rate','hand_acc_16g_x','hand_acc_16g_y','hand_acc_16g_z']] #TRY ONLY WITH ACC HAND+HEART RATE
#data = data[['time','activity','hand_temp','hand_acc_16g_x','hand_acc_16g_y','hand_acc_16g_z']] #TRY ONLY WITH HAND MEASUREMENTS
data = data[[
'time', 'activity', 'heart_rate', 'hand_temp', 'chest_temp', 'ankle_temp'
]]
""" Choose which configuration to use by commenting in."""
### This mode uses summary statistics, rolling mean and std.
#MODE = 'summaries'
#wsizes = [10,20,40,80,120]
### This mode allows prediction using full data for a given window.
MODE = 'windows'
#!/usr/bin/env python3
# -*- coding: utf-8 -*-
"""
Created on Sat Nov 18 13:56:16 2017
@author: rv
"""
import numpy as np
from matplotlib import pyplot as plt
from exploration import loadSubject
subs = [101, 102, 105, 108]
data = [loadSubject(i, interpolate=False) for i in subs]
acts = []
for _, d in enumerate(data):
for u in d['activity'].unique():
if u not in acts:
acts.append(u)
m_total = []
s_total = []
for _, d in enumerate(data):
act_desc = []
for act in acts:
data_act = d.loc[d['activity'] == act]
desc_ = data_act.describe().T
### This mode uses summary statistics, rolling mean and std.
#MODE = 'summaries'
wsize = 120
nEsts = 25
"""----------------------------------------------------------------------------
-------------------------------------------------------------------------------
----------------------------------------------------------------------------"""
Results = []
for trainSub in Subs:
print('Training Subject '+str(trainSub))
testSubs = Subs[:]
testSubs.remove(trainSub)
data = loadSubject(trainSub)
data = data[features]
data_w = summarizeWindows(data,wsize)
#Dropping time as a prediction parameter
data_w = data_w.drop('time',axis=1)
data_w = data_w.iloc[np.random.permutation(len(data_w))]
X = data_w.drop('activity',axis=1)
y = data_w['activity']
clf = RandomForestClassifier(n_estimators=nEsts)
clf.fit(X,y)
resSub = []
for sub in Subs:
print('Testing Subject '+str(sub))
wsize = 120
nEsts = 25
"""----------------------------------------------------------------------------
-------------------------------------------------------------------------------
----------------------------------------------------------------------------"""
Results = []
for testSub in Subs:
print('Test Subject ' + str(testSub))
trainSubs = Subs[:]
trainSubs.remove(testSub)
init = 1
for sub in trainSubs:
print('Loading Subject ' + str(sub))
data = loadSubject(sub)
# data = data[features]
data_w = summarizeWindows(data, wsize)
#Dropping time as a prediction parameter
data_w = data_w.drop('time', axis=1)
data_w = data_w.iloc[np.random.permutation(len(data_w))]
X_ = data_w.drop('activity', axis=1)
y_ = data_w['activity']
if init == 1:
X_train = X_
y_train = y_
init = 0