def save_key_attributes():
if len(event_log) > 0:
selected_caseid = caseid_value_chosen.get()
selected_eventid = eventid_value_chosen.get()
selected_activity = activity_value_chosen.get()
selected_timestamp = timestamp_value_chosen.get()
global extracted_data #event_log only with key attributes
if selected_eventid=="-- no attribute --" :
extracted_data = event_log[
[selected_caseid, selected_activity, selected_timestamp]]
extracted_data["Event"] = list(range(0,len(event_log.index)))
cols = extracted_data.columns.tolist()
cols = cols[:1]+ cols[-1:] + cols[1:3]
extracted_data = extracted_data[cols]
if selected_eventid!="-- no attribute --" :
extracted_data = event_log[[selected_caseid, selected_eventid, selected_activity, selected_timestamp]]
extracted_data.columns = \
["Case", "Event", "Activity", "Timestamp"]
extracted_data = extracted_data.dropna(subset=['Case'])
PageTwo.firstpreprocess = extracted_data
extracted_data = extracted_data.sort_values(["Case", "Timestamp"],
ascending=[True, True])
form_dict = {"2020-01-02 03:04:05.006":"%Y-%m-%d %H:%M:%S.%f",
"2020-01-02 03:04:05":"%Y-%m-%d %H:%M:%S",
"2020-01-02 03:04:05.006 PM":"%Y-%m-%d %I:%M:%S.%f %p",
"2020-01-02 03:04:05 PM":"%Y-%m-%d %I:%M:%S %p",
"20-01-02 03:04:05.006":"%y-%m-%d %H:%M:%S.%f",
"20-01-02 03:04:05":"%y-%m-%d %H:%M:%S",
"20-01-02 03:04:05.006 PM":"%y-%m-%d %I:%M:%S.%f %p",
"20-01-02 03:04:05 PM":"%y-%m-%d %I:%M:%S %p",
"2020/01/02 03:04:05.006": "%Y/%m/%d %H:%M:%S.%f",
"2020/01/02 03:04:05": "%Y/%m/%d %H:%M:%S",
"2020/01/02 03:04:05.006 PM": "%Y/%m/%d %I:%M:%S.%f %p",
"2020/01/02 03:04:05 PM": "%Y/%m/%d %I:%M:%S %p",
"20/01/02 03:04:05.006": "%y/%m/%d %H:%M:%S.%f",
"20/01/02 03:04:05": "%y/%m/%d %H:%M:%S",
"20/01/02 03:04:05.006 PM": "%y/%m/%d %I:%M:%S.%f %p",
"20/01/02 03:04:05 PM": "%y/%m/%d %I:%M:%S %p"
}
if timeform_value_chosen.get() not in form_dict.keys() :
form = timeform_value_chosen.get()
if iserror(dt.strptime, extracted_data['Timestamp'][0], form):
if "\"" in form or "\'" in form:
words = ["Type timestamp format without", "\'", "or", "\""]
words = " ".join(words)
messagebox.showinfo("Error", words)
else: messagebox.showinfo("Error", "Wrong input for timestamp format: use format in 'datetime' packages. (ex: %Y-%m-%d %H:%M:%S)")
else:
time = extracted_data['Timestamp'].apply(lambda x: dt.strptime(x, form))
unixtime = time.apply(lambda x: (x - dt(1970, 1, 1)).total_seconds())
extracted_data['Timestamp'] = time
extracted_data['unixtime'] = unixtime
PageTwo.extracted_data = extracted_data
messagebox.showinfo("Message", "Loaded")
tk.Frame.grid_forget()
parent.show_frame1()
if timeform_value_chosen.get() in form_dict.keys():
form = form_dict[timeform_value_chosen.get()]
if iserror(dt.strptime, extracted_data['Timestamp'][0], form):
messagebox.showinfo("Error",
"{0} is not matched with {1}".format(extracted_data['Timestamp'][0], form))
else:
time = extracted_data['Timestamp'].apply(lambda x: dt.strptime(x, form))
unixtime = time.apply(lambda x: (x - dt(1970, 1, 1)).total_seconds())
extracted_data['Timestamp'] = time
extracted_data['unixtime'] = unixtime
PageTwo.extracted_data = extracted_data
messagebox.showinfo("Message", "Loaded")
parent.show_frame1()
else:
messagebox.showinfo("Error", "Error")
s = '00'
bp = h + ':' + m + ':' + s
return bp
train = pd.read_csv("train.csv")
mp = pd.read_csv("train.csv")["duration"]
time = mp.apply(checki)
def func_sec(time_string):
h, m, s = time_string.split(':')
return int(h) * 3600 + int(m) * 60 + int(s)
time1 = time.apply(func_sec)
data_train["duration"] = time1
data_train.head()
# Visualization # Individual Plots
plt.hist(data_train["category"])
plt.show()
plt.plot(data_train["adview"])
plt.show()
# Remove videos with adview greater than 2000000 as outlier
data_train = data_train[data_train["adview"] < 2000000]
# Heatmap
import seaborn as sns
def convert_time(time): time = time.apply(_parse_time) time = time.apply(_convert_delta_to_hours,args=(time[0],)) #time = time.round(2) return time