def grid_search_run(train_dfs, label_dfs, test_dfs, sax_words_as_svec, window_list, nbin_list, alphabet_list, verbose):
dask_output = []
i = 1
for train_df, label_df, test_df in zip(train_dfs, label_dfs, test_dfs):
for window in window_list:
for nbins in nbin_list:
for alphabet_size in alphabet_list:
s_time = gt.time()
if verbose:
print('dset: {}\tstime: {}\tw: {}\tnb: {}\ta: {}'.
format(i, gt.time(), window, nbins, alphabet_size))
alphabet = string.ascii_uppercase[:alphabet_size]
sax = ToSAX(window, nbins=nbins, alphabet=alphabet)
train_sax_df = run_sax_along_df(train_df, sax, sax_words_as_svec, verbose)
test_sax_df = run_sax_along_df(test_df, sax, sax_words_as_svec, verbose)
print_toSAX_csv(train_sax_df, i, window, nbins, alphabet_size, 'train')
print_toSAX_csv(test_sax_df, i, window, nbins, alphabet_size, 'test')
# train_sax_df = delayed(run_sax_along_df)(train_df, sax, sax_words_as_svec, verbose)
# test_sax_df = delayed(run_sax_along_df)(test_df, sax, sax_words_as_svec, verbose)
# dask_output.append(delayed(print_SAX_to_csv)(train_sax_df, window, nbins, alphabet_size, 'train'))
# dask_output.append(delayed(print_SAX_to_csv)(test_sax_df, window, nbins, alphabet_size, 'test'))
if verbose:
print('SAXed dataset: {}\tstart time: {}\tend time: {}'.format(i, s_time, gt.time()))
i += 1
def main():
print(os.listdir())
time = gt.GetTime()
verbose = True
full_run = True
grid_search = True
sax_words_as_string_vec = True # scikit learn requires string vectors, does not accept 2d list
#----------SAX--------------#
window = 4
stride = window
nbins = 5
alphabet_size = 5
alphabet = string.ascii_uppercase[:alphabet_size]
sax = ToSAX(window, stride, nbins, alphabet)
#-----------GRID------------#
window_list = range(5, 30, 5)
nbin_list = range(5, 12, 2)
alphabet_list = range(4, 9, 2)
if verbose:
print(time.initial_time)
if full_run:
test_fns, train_fns, labels_fns = get_fns(verbose)
# dfs below are generators
train_dfs = get_dataframes(verbose, train_fns)
label_dfs = get_dataframes(verbose, labels_fns)
test_dfs = get_dataframes(verbose, test_fns)
if grid_search:
if verbose: print('doing grid search')
grid_search_run(train_dfs, label_dfs, test_dfs, sax_words_as_string_vec, window_list, nbin_list, alphabet_list, verbose)
else:
regular_run(train_dfs, label_dfs, test_dfs, sax, sax_words_as_string_vec, window, nbins, alphabet_size, verbose)
else:
if verbose: print('in else')
i = 1
dataset1_test = 'hw1_datasets/dataset4/test_normalized.csv'
test1_labels = 'hw1_datasets/dataset4/test_labels.csv'
dataset1_train = 'hw1_datasets/dataset4/train_normalized.csv'
dataset1_train_labels = 'hw1_datasets/dataset4/train_labels.csv'
train1 = pd.read_csv(dataset1_train, index_col=0)
test1 = pd.read_csv(dataset1_test, index_col=0)
train1_labels = pd.read_csv(dataset1_train_labels, index_col=0)
# ts_sax_df = run_sax_along_df(train1.head(n=10), sax, sax_words_as_string_vec, verbose)
train_sax_df = run_sax_along_df(train1, sax, sax_words_as_string_vec, verbose)
test_sax_df = run_sax_along_df(test1, sax, sax_words_as_string_vec, verbose)
if verbose:
print('Results Found')
print('sax df shape: {}'.format(train_sax_df.shape))
# print(train_sax_df)
print_toSAX_csv(train_sax_df, i, window, nbins, alphabet_size, 'train', True)
print_toSAX_csv(test_sax_df, i, window, nbins, alphabet_size, 'test', True)
if verbose:
print('-------Completed!-------')
print('Started at: {}\tFinished at: {}'.format(time.initial_time, gt.time()))
def search_time(self):
print(u"请确保将查询的人员名单写入name_total.txt，并保证其在运行目录下")
print(u"若人员较多则所需时间较长，请耐心等待...")
GetID.get_id()
GetTime.get_time()
print(u"详细时长已保存在time_total.txt")
print(u"按任意键返回主菜单..."),
raw_input()
self.__init__()
def print_results_to_csv(predictions, dataset_num, sax, s_time, clf='kNN'):
print('Printing Results')
test_output_fn = 'test_output/results_dataset{}_{}_w{}_nb{}_a{}_s{}_e{}.csv'.format(
dataset_num, clf, sax.window, sax.nbins, sax.alphabet_cardinality, s_time, gt.time())
with open(test_output_fn, 'w') as results:
for y in predictions:
results.write('{0}\n'.format(y))
def main(fname):
'''主程序'''
print('提示：如果直接回车，要转换的文字将为“要转换的文字.txt”中的文字')
ms = input('输入要转换成图片的文字: \n ')
print('开始时间：', GetTime.getTime())
if ms == '':
try:
f = open('要转换的文字.txt', 'r', encoding='utf-8')
ms = str(f.readlines())
f.close()
except:
print('缺少默认文件')
print('len(ms)=', len(ms))
print('start', GetTime.getTime())
saveimg(fname, str_to_arr(ms))
print('结束时间：', GetTime.getTime())
input('按任意键继续')
def update_results():
full_filename = os.path.join(app.config['UPLOAD_FOLDER'],
'WashWatch_blau.png')
print(full_filename)
#The next line needs to be uncommented, when a camera is adapted!
#pic.takePicture()
time_left = time.getTime()
return render_template('index.html',
user_image=full_filename,
time=time_left)
def regular_run(train_dfs, label_dfs, test_dfs, sax, sax_words_as_string_vec, window, nbins, alphabet_size, verbose):
i=1
for train_df, label_df, test_df in zip(train_dfs, label_dfs, test_dfs):
s_time = gt.time()
print(label_df.shape)
# print(label_df)
train_sax_df = run_sax_along_df(train_df, sax, sax_words_as_string_vec, verbose)
test_sax_df = run_sax_along_df(test_df, sax, sax_words_as_string_vec, verbose)
print_toSAX_csv(train_sax_df, i, window, nbins, alphabet_size, 'train')
print_toSAX_csv(test_sax_df, i, window, nbins, alphabet_size, 'test')
if verbose:
print('SAXed dataset: {}\tstart time: {}\tend time: {}'.format(i, s_time, gt.time()))
print('sax df shape: {}'.format(train_sax_df.shape))
i += 1
def CreateAbnormalErrorSet(userName, apiToken):
minutesOfConsistentErrors = random.randint(15, 90)
overallErrorClass = random.randint(1, 4)
elementInErrorClassArray = random.randint(0, 4)
for i in range(minutesOfConsistentErrors):
createdAt = GetTime.getTime()
avgNumberOfErrorsPerMinute = random.randint(10, 25)
for j in range(avgNumberOfErrorsPerMinute):
eventDict = se.postEvent(userName, apiToken, overallErrorClass, elementInErrorClassArray, createdAt)
sm.sendMeasurements(userName, apiToken, eventDict, avgNumberOfErrorsPerMinute, createdAt)
# Commented out all the multi-threading during inital stages to make sure everything is working
#
#
#
#
#
# while numberOfEventsSent <= minutesOfConsistentErrors:
#
#
#
# timeofLastAbnormalError = time.time()
# abnormalErrorSetInterval = random.uniform(1,3)
# numberOfEventsSent += 1
# if time.time() >= (timeofLastAbnormalError + abnormalErrorSetInterval):
# createdAt = GetTime.getTime()
# timeofLastAbnormalError = time.time()
# abnormalErrorSetInterval = random.uniform(1,3)
# eventDict = se.postEvent(userName, apiToken, overallErrorClass, elementInErrorClassArray, createdAt)
# numberOfEventsSent += 1
#
# sm.sendMeasurements(userName, apiToken, eventDict, avgNumberOfErrorsPerMinute, createdAt)
# _thread.exit()
def main():
#----------MAIN-------------#
print(os.listdir())
time = gt.GetTime()
verbose = True
full_run = True
#----------SAX--------------#
window = 4
stride = window
nbins = 5
alphabet_size = 5
alphabet = string.ascii_uppercase[:alphabet_size]
sax = ToSAX(window, stride, nbins, alphabet)
#-----------CLF------------#
# grid_search = True
k = 1
njobs = cpu_count() - 1
if verbose:
print(time.initial_time)
if full_run:
if verbose: print('in full run')
train_fns, test_fns, train_label_fns, test_label_fns = get_fns(sax)
# dfs below are generators
train_dfs = get_dataframes(train_fns)
train_label_dfs = get_dataframes(train_label_fns, True)
test_label_dfs = get_dataframes(test_label_fns, True)
test_dfs = get_dataframes(test_fns)
results_array = []
i = 1
for train_sax, train_labels, test_sax, test_labels in zip(
train_dfs, train_label_dfs, test_dfs, test_label_dfs):
s_time = gt.time()
print(test_labels)
# create a Bag of patterns and transform sax representations to a tfidf matrix
BoP = BagOfPatterns(train_sax, train_labels, test_sax, test_labels,
sax)
BoP.fit(sax_letter_cardinality=sax.alphabet_cardinality,
do_tfidf=True,
ngram_len=2,
min_term_freq=1,
max_term_freq=1)
train_tfidf, test_tfidf = BoP.transform()
# classify tfidfs using kNN
kNN = KNeighborsClassifier(n_neighbors=k,
weights='uniform',
algorithm='auto',
metric='minkowski',
p=2,
n_jobs=njobs)
kNN.fit(train_tfidf, train_labels)
test_predicted = kNN.predict(test_tfidf)
print_results_to_csv(test_predicted, i, sax, time.initial_time,
'kNN')
if verbose:
print('Results Found for dataset: {}\ttime: {}'.format(
i, gt.time()))
i += 1
else:
print('in else')
i = 1
data_test_fn = 'hw1_datasets/dataset{}/test_sax_w{}_nb{}_v{}.csv'.format(
i, window, nbins, alphabet_size)
data_train_fn = 'hw1_datasets/dataset{}/train_sax_w{}_nb{}_v{}.csv'.format(
i, window, nbins, alphabet_size)
test_labels = 'hw1_datasets/dataset{}/test_labels.csv'.format(i)
dataset_train_labels = 'hw1_datasets/dataset{}/train_labels.csv'.format(
i)
dataset_test_labels = 'hw1_datasets/dataset{}/test_labels.csv'.format(
i)
x_sax = pd.read_csv(data_train_fn, index_col=0)
test_sax = pd.read_csv(data_test_fn, index_col=0)
# converting from pandas to list for scikit learn
train_labels = pd.read_csv(dataset_train_labels,
index_col=0).transpose().values.tolist()[0]
test_labels = pd.read_csv(dataset_test_labels,
index_col=0).transpose().values.tolist()[0]
# create a Bag of patterns and transform sax representations to a tfidf matrix
BoP = BagOfPatterns(x_sax, train_labels, test_sax, test_labels, sax)
BoP.fit(sax_letter_cardinality=sax.alphabet_cardinality,
do_tfidf=True,
ngram_len=2,
min_term_freq=1,
max_term_freq=1)
train_tfidf, test_tfidf = BoP.transform()
# classify tfidfs using kNN
kNN = KNeighborsClassifier(n_neighbors=k,
weights='uniform',
algorithm='auto',
metric='minkowski',
p=2,
n_jobs=njobs)
kNN.fit(train_tfidf, train_labels)
test_predicted = kNN.predict(test_tfidf)
print_results_to_csv(test_predicted, i, sax, time.initial_time, 'kNN')
if verbose: print('Results Found')
if verbose:
print('-------Completed!{}-------')
print('Started at: {}\tFinished at: {}'.format(time.initial_time,
gt.time()))
apiToken = args.api
pollingFrequency = args.freq
# Metadata definition for the metrics, this will create the metric definitions
# if they do not exist
# dm.defineAllMetrics(userName, apiToken)
NON_CONSISTENT_ERROR_TYPE = 0
# Live action - push random errors in an ongoing basis
# at random intervals, also start sending consistent error messages (still with the random errors at the same speed)
counter = 0
total_error_counter = 0
while True:
createdAt = GetTime.getTime()
numberOfErrors = random.randint(1, 4)
for i in range(numberOfErrors):
eventDict = se.postEvent(userName, apiToken, NON_CONSISTENT_ERROR_TYPE,
NON_CONSISTENT_ERROR_TYPE, createdAt)
sm.sendMeasurements(userName, apiToken, eventDict, numberOfErrors,
createdAt)
print(str(counter))
counter += 1
# Commented out the abnormal issues section of the code during testing to make sure normal events hit the API
#
#
#
#
#
def get_dataframes(verbose, fn_arr):
if verbose:
print('In get_dataframes: {}'.format(gt.time()))
file_generator = (pd.read_csv(fn, index_col=0) for fn in fn_arr)
return file_generator
def find_prediction_metrics(dtw_list, k_list, dist='DTW', verbose=True):
# Grid search crashed ONLY GOT W=5 PREDICTIONS for w in w_list:
if dist == 'DTW':
print('in DTW')
for dtw_w in dtw_list:
for k in k_list:
for i in range(1, 6): #each dataset
try:
if verbose:
print('TRY - dset: {}\ttime: {}\tw: {}\tk: {}'.
format(i, gt.time(), dtw_w, k))
y = get_y(i)
yhat = get_class_predictions(i, dist, dtw_w, k)
f1 = f1_score(
y, yhat, average=None
) # returns list score [pos neg], can use weighted
f1_avg = f1_score(
y, yhat, average='weighted'
) # returns list score [pos neg], can use weighted
acc = accuracy_score(y, yhat)
class_counts = yhat[0].value_counts(sort=False)
save_kNN_metrics(i, dist, dtw_w, k, acc, f1, f1_avg,
class_counts)
i += 1
except:
print('--------------')
print('These parameters were skipped:')
print(
'EXCEPT - dset: {}\ttime: {}\tw: {}\tk: {}'.format(
i, gt.time(), dtw_w, k))
print('test_output/test_results_dataset{}_{}_k{}.csv'.
format(i, dist, k))
print('--------------')
continue
else:
print('in EUC')
for k in k_list:
for i in range(1, 6): #each dataset
try:
if verbose:
print('TRY - dset: {}\ttime: {}\tk: {}'.format(
i, gt.time(), k))
y = get_y(i)
yhat = get_class_predictions(i, dist, None, k)
f1 = f1_score(
y, yhat, average=None
) # returns list score [pos neg], can use weighted
f1_avg = f1_score(
y, yhat, average='weighted'
) # returns list score [pos neg], can use weighted
acc = accuracy_score(y, yhat)
class_counts = yhat[0].value_counts(sort=False)
save_kNN_metrics(i, dist, np.nan, k, acc, f1, f1_avg,
class_counts)
i += 1
except:
print('--------------')
print('These parameters were skipped:')
print('EXCEPT - dset: {}\ttime: {}\tk: {}'.format(
i, gt.time(), k))
print(
'test_output/test_results_dataset{}_{}_k{}.csv'.format(
i, dist, k))
print('--------------')
continue
#print(len(图片数组), type(图片数组))
#print(图片数组)
mpimg.imsave(fname, 图片数组)
def main(fname):
'''主程序'''
print('提示：如果直接回车，要转换的文字将为“要转换的文字.txt”中的文字')
ms = input('输入要转换成图片的文字: \n ')
print('开始时间：', GetTime.getTime())
if ms == '':
try:
f = open('要转换的文字.txt', 'r', encoding='utf-8')
ms = str(f.readlines())
f.close()
except:
print('缺少默认文件')
print('len(ms)=', len(ms))
print('start', GetTime.getTime())
saveimg(fname, str_to_arr(ms))
print('结束时间：', GetTime.getTime())
input('按任意键继续')
if __name__ == '__main__':
#for i in range(9):
filename = GetTime.getTime(2) + '.png'
main(filename)
print(filename, 'is ok')
print('end', '--' * 10)
@LastModifiedBy: ywyz
@Github: https://github.com/ywyz
@LastEditors: ywyz
@LastEditTime: 2020-07-22 22:02:53
'''
import openpyxl
import GetTime
import datetime
import ProcessingForm
import CompareTime
import CopyCell
current_time = datetime.datetime.now()
# 获取当前时间
time = GetTime.GetTime(current_time)
time_strings = time.ReturnTimeStrings()
# 获取当前时间字符串
print("今日时间：", time_strings)
namestrings = time.TimeStrings()
names = input("输入文件名字或者拖动文件到这儿来：")
wb = openpyxl.load_workbook(names) # 打开文件
wb_sheet = wb["Sheet1"] # 打开工作簿1
# 获取表格最多行数
processForm = ProcessingForm.ProcessingForm(wb_sheet)
max_row = processForm.GetMaxROW()
# 比较时间，删除过时的行
time = CompareTime.CompareTime(wb_sheet, time_strings, max_row)
oled.text('Well Connected', 5, 20) #显示连接是否成功及IP地址
oled.text("IP:" + wlan.ifconfig()[0], 0, 35)
oled.show()
utime.sleep(1)
rtc = RTC()
update_time = utime.ticks_ms() - First_time
print(update_time)
while True:
if not wlan.isconnected():
machine.reset()
elif utime.ticks_ms() - update_time >= First_time:
Year, Month, Day, Hour, Week, Minute, Second = GetTime.Get_Time(
) #通过GetTime文件爬取时间
rtc.datetime(
(Year, Month, Day, Week, Hour, Minute, Second, 0)) #使用RTC实时时钟记录时间
#date = "{:02}-{:02}-{:02}".format(rtc.datetime()[0], rtc.datetime()[1], rtc.datetime()[2])
#time = "{:02}:{:02}:{:02}".format(rtc.datetime()[4], rtc.datetime()[5], rtc.datetime()[6])
Week = Chance.Week_English(rtc.datetime()[3]) #将数字星期变换成英文缩写
Temperature_Now, Temperature_Min, Temperature_Max, Weather = GetWeather.Get_Weather(
)
Weather = Chance.Weather_English(Weather)
update_time = utime.ticks_ms()
else:
update_time = utime.ticks_ms() - First_time + Second_time