#import pandas as pd
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import confusion_matrix, roc_curve, auc
from sklearn.svm import SVC
import matplotlib.pyplot as plt
from sklearn.metrics import accuracy_score
#%% get data
data_df = utils.load_object('faps_for_train.pkl')
valence = utils.load_object('valence.pkl')
arousals = utils.load_object('arousal.pkl')
valence_list = valence['valence'].tolist()
arousal_list = arousals['arousal'].tolist()
data_df = pu.match_label_with_sample(data_df, valence_list)
data_df = pu.match_label_with_sample(data_df, arousal_list, col_name='arousal')
label = data_df['label'].values.astype(np.int64)
data = data_df.drop(columns=['ori_idx', 'label', 'arousal']).values.astype(
np.float32)
#test_df = utils.load_object('valence_for_train.pkl')
#test_df = pu.match_label_with_sample(test_df,valence_list)
#
#label_test = test_df['label'].values.astype(np.int64)
#data_test = test_df[['slope_qr','delta_pq','delta_qr']].values.astype(np.float32)
# split train test data
#X_train , X_test, y_train, y_test = train_test_split(data,label,test_size=0.1)
X_train, X_test, y_train, y_test = train_test_split(data,
cufflinks.go_offline(connected=True)
init_notebook_mode(connected=True)
#%%
from sklearn.preprocessing import StandardScaler
#%% get data
import utils
import preprocessing.pre_utils as pu
data_df = utils.load_object('faps_for_train.pkl')
valence = utils.load_object('valence.pkl')
#arousals = utils.load_object('arousal.pkl')
valence_list = valence['valence'].tolist()
#arousal_list = arousals['arousal'].tolist()
data_df = pu.match_label_with_sample(data_df, valence_list)
#data_df = pu.match_label_with_sample(data_df,arousal_list,col_name='arousal')
#%% plot scatter matrix
fig = ff.create_scatterplotmatrix(data_df[[0, 1, 12, 13, 'label']],
diag='histogram',
index='label',
height=1000,
width=1000)
plotly.offline.plot(fig)
#%%
##iaps_class = iaps(r"C:\Users\DSPLab\Research\affective-monitor-model\preprocessing\IAPSinfoFile_Final.txt")
iaps_class = iaps(r"E:\Research\affective-monitor-model\preprocessing")
sample_list_from_pic_id = iaps_class.get_sample_idx(2141)
import numpy as np
import preprocessing.pre_utils as pu
#import pandas as pd
from sklearn.model_selection import train_test_split, GridSearchCV
from sklearn.preprocessing import StandardScaler
from sklearn.metrics import confusion_matrix, roc_curve, auc
from sklearn.svm import SVC
import matplotlib.pyplot as plt
from sklearn.metrics import accuracy_score
#%% get data
data_df = utils.load_object('pd_for_train.pkl')
arousals = utils.load_object('arousal.pkl')
arousals_list = arousals['arousal'].tolist()
data_df = pu.match_label_with_sample(data_df, arousals_list)
label = data_df['label'].values.astype(np.int64)
data = data_df[['slope_qr', 'delta_pq', 'delta_qr']].values.astype(np.float32)
test_df = utils.load_object('pd_for_test.pkl')
test_df = pu.match_label_with_sample(test_df, arousals_list)
label_test = test_df['label'].values.astype(np.int64)
data_test = test_df[['slope_qr', 'delta_pq',
'delta_qr']].values.astype(np.float32)
# split train test data
#X_train , X_test, y_train, y_test = train_test_split(data,label,test_size=0.1)
X_train, X_test, y_train, y_test = train_test_split(data,
label,
#%% get data for valence
import utils
import preprocessing.pre_utils as pu
import numpy as np
data_df = utils.load_object('faps_for_train.pkl')
valence = utils.load_object('valence.pkl')
valence_list = valence['valence'].tolist()
data_df = pu.match_label_with_sample(data_df,valence_list)
data_df = data_df.drop_duplicates('ori_idx')
label = data_df['label'].values.astype(np.int64)
data = data_df.drop(columns=['ori_idx','label']).values.astype(np.float32)
ori_idx = data_df['ori_idx'].tolist()
#X_train , X_test, y_train, y_test = train_test_split(data,label,test_size=0,random_state=42)
#%% get model
from sklearn.externals import joblib
from sklearn.metrics import confusion_matrix, accuracy_score
import pandas as pd
filename = 'save_model/valence_svm.pkl'
loaded_model = joblib.load(filename)
valence_predicted = loaded_model.predict(data).tolist()
print(confusion_matrix(label, valence_predicted))
print('accuracy: {:.2f}'.format(accuracy_score(label, valence_predicted)))
val_np = np.array([valence_predicted,ori_idx]).transpose()
pred_val_df = pd.DataFrame(val_np,columns=['valence','ori_idx'])
#%% metrics
#%% visualize pqr
#ppd.plot_pqr_slideshow(pd_pqr_df,51,smooth=True)
#%% data selection
#sel_pic_list = utils.load_object('selected_idx_list.pkl')
#pd_pqr_df = pd_pqr_df[pd_pqr_df['ori_idx'].isin(sel_pic_list)]
#%% data selection
import preprocessing.pre_utils as pu
import pandas as pd
arousals = utils.load_object('arousal.pkl')
arousals_list = arousals['arousal'].tolist()
pd_sel_df = pd_pqr_df.copy()
pd_sel_df = pu.match_label_with_sample(pd_sel_df, arousals_list)
pd_ar_df = pd_sel_df[((pd_sel_df['label'] == 1) & (pd_sel_df['area_ql'] > 20))]
pd_nar_df = pd_sel_df[((pd_sel_df['label'] == 2) &
(pd_sel_df['area_ql'] <= 20))]
pd_nar_df = pd_nar_df.sample(pd_ar_df.shape[0])
samples = pd.concat([pd_ar_df, pd_nar_df], ignore_index=True)
samples = samples.sample(frac=1)
# plot slide show of data
#ppd.plot_pqr_slideshow(samples,'all',smooth=False,label=samples['label'].tolist())
# drop label before save it to pickle
samples = samples.drop('label', axis=1)