def read_feature_6(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(
new_df.Date.dt.month
) # Problem Solved By link:- https://www.shanelynn.ie/summarising-aggregation-and-grouping-data-in-python-pandas/
months_available = groups.groups.keys(
) # https://stackoverflow.com/questions/28844535/python-pandas-groupby-get-list-of-groups
feat = read_feature_4(path)
M_observed = []
for month in months_available:
M_observed.append(groups.get_group(month)["Magnitude"].max())
M_expected = []
for a, b in feat:
if (b == 0.0 or a == 0.0):
M_expected.append(0.0)
else:
M_expected.append(a / b)
deficit = np.subtract(np.array(M_observed), np.array(M_expected))
# print(deficit)
return deficit
def read_feature_2(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
# x = new_df['Date']+' '+new_df['Time']
# new_df["Datetime"] = pd.to_datetime(x ,format='%Y/%m/%d %H:%M:%S.%f')
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(new_df.Date.dt.month)
#print(groups['Magnitude'].mean())
return groups['Magnitude'].mean()
def read_feature_2(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
# x = new_df['Date']+' '+new_df['Time']
# new_df["Datetime"] = pd.to_datetime(x ,format='%Y/%m/%d %H:%M:%S.%f')
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(
new_df.Date.dt.month
) # Problem Solved By link:- https://www.shanelynn.ie/summarising-aggregation-and-grouping-data-in-python-pandas/
#print(groups['Magnitude'].mean())
return groups['Magnitude'].mean()
def read_feature_5(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(
new_df.Date.dt.month
) # Problem Solved By link:- https://www.shanelynn.ie/summarising-aggregation-and-grouping-data-in-python-pandas/
months_available = groups.groups.keys(
) # https://stackoverflow.com/questions/28844535/python-pandas-groupby-get-list-of-groups
lst = []
for month in months_available:
sorted_mag = groups.get_group(month)["Magnitude"]
sorted_mag = np.sort(sorted_mag)
n = len(sorted_mag)
mapped_mag = dict(Counter(sorted_mag))
M = []
N = []
for mag in mapped_mag:
M.append(mag)
N.append(n)
n = n - mapped_mag[mag]
N = list(map(math.log, N))
n = len(sorted_mag)
part1 = np.sum(np.array(M) * np.array(N))
part2 = np.sum(np.array(M)) * np.sum(np.array(N))
part3 = np.square(np.sum(np.array(M)))
part4 = np.sum(np.square(np.array(M)))
b = 0.0
eta = 0.0
if n == 1:
b = 0.0
a = np.sum(np.array(N)) / n
eta = 0.0
else:
b = (n * part1 - part2) / (part3 - n * part4)
M = b * np.array(M)
a = np.sum(np.array(N) + np.array(M)) / n
eta = np.sum(np.square(np.array(N) - (a - M))) / (n - 1)
lst.append(eta)
# print(lst)
return np.array(lst)
def read_feature_1(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date'] + ' ' + new_df['Time']
new_df["Datetime"] = pd.to_datetime(x, format='%Y/%m/%d %H:%M:%S.%f')
# x = new_df['Date']
# new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(new_df.Datetime.dt.month).agg({'Datetime': ['first',
'last']}) # Problem Solved By link:- https://www.shanelynn.ie/summarising-aggregation-and-grouping-data-in-python-pandas/
days = (groups.Datetime['last'] - groups.Datetime['first'])
groups['seconds'] = days / np.timedelta64(1, 's')
groups['days'] = days
#print(days.dt.days)
return np.float64(days.dt.days)
def read_feature_1(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date'] + ' ' + new_df['Time']
new_df["Datetime"] = pd.to_datetime(x, format='%Y/%m/%d %H:%M:%S.%f')
# x = new_df['Date']
# new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(new_df.Datetime.dt.month).agg(
{'Datetime': ['first', 'last']})
days = (groups.Datetime['last'] - groups.Datetime['first'])
groups['seconds'] = days / np.timedelta64(1, 's')
groups['days'] = days
#print(days.dt.days)
return np.float64(days.dt.days)
def read_feature_5(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(
new_df.Date.dt.month)
months_available = groups.groups.keys()
lst = []
for month in months_available:
sorted_mag = groups.get_group(month)["Magnitude"]
sorted_mag = np.sort(sorted_mag)
n = len(sorted_mag)
mapped_mag = dict(Counter(sorted_mag))
M = []
N = []
for mag in mapped_mag:
M.append(mag)
N.append(n)
n = n - mapped_mag[mag]
N = list(map(math.log, N))
n = len(sorted_mag)
part1 = np.sum(np.array(M) * np.array(N))
part2 = np.sum(np.array(M)) * np.sum(np.array(N))
part3 = np.square(np.sum(np.array(M)))
part4 = np.sum(np.square(np.array(M)))
b = 0.0
eta = 0.0
if n == 1:
b = 0.0
a = np.sum(np.array(N)) / n
eta = 0.0
else:
b = (n * part1 - part2) / (part3 - n * part4)
M = b * np.array(M)
a = np.sum(np.array(N) + np.array(M)) / n
eta = np.sum(np.square(np.array(N) - (a - M))) / (n - 1)
lst.append(eta)
# print(lst)
return np.array(lst)
def read_feature_3(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(new_df.Date.dt.month)
T = read_feature_1(path)
months_available = groups.groups.keys()
dE = []
for month in months_available:
dE.append(
np.sum(
groups.get_group(month)["Magnitude"].apply(
lambda x: np.sqrt(10**(11.8 + 1.5 * x)))))
dE = np.array(dE)
old = np.seterr('ignore') # used to ignore divide by zero exception
return np.array(dE / T)
def read_feature_6(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(new_df.Date.dt.month)
months_available = groups.groups.keys()
feat = read_feature_4(path)
M_observed = []
for month in months_available:
M_observed.append(groups.get_group(month)["Magnitude"].max())
M_expected = []
for a, b in feat:
if (b == 0.0 or a == 0.0):
M_expected.append(0.0)
else:
M_expected.append(a / b)
deficit = np.subtract(np.array(M_observed), np.array(M_expected))
# print(deficit)
return deficit
def read_feature_3(path):
df = read_data(path)
df['Magnitude'] = df['Magnitude'].apply(pd.to_numeric)
new_df = df.loc[df["Magnitude"] > threshold]
x = new_df['Date']
new_df["Date"] = pd.to_datetime(x, format='%Y/%m/%d')
del new_df["Time"]
groups = new_df.groupby(
new_df.Date.dt.month
) # Problem Solved By link:- https://www.shanelynn.ie/summarising-aggregation-and-grouping-data-in-python-pandas/
T = read_feature_1(path)
months_available = groups.groups.keys(
) # https://stackoverflow.com/questions/28844535/python-pandas-groupby-get-list-of-groups
dE = []
# Solved using https://pandas.pydata.org/pandas-docs/stable/groupby.html - (get_group() attribute)
for month in months_available:
dE.append(
np.sum(
groups.get_group(month)["Magnitude"].apply(
lambda x: np.sqrt(10**(11.8 + 1.5 * x)))))
dE = np.array(dE)
old = np.seterr('ignore') # used to ignore divide by zero exception
return np.array(dE / T)