def display_ij(self,n_dimensions,X_fs,X_val,Y_val,X_test,Y_test,i,j,classifieur_model):
ACP = PCA(n_components=n_dimensions)
ACP.fit(X_fs)
X_valr = ACP.tranform(X_val)
X_testr = ACP.transform(X_test)
index_min = [min(X_valr[:,i].min(),X_testr[:,i].min()) - .5 for i in range(n_dimensions)]
index_max = [max(X_valr[:,i].max(),X_testr[:,i].max()) + .5 for i in range(n_dimensions)]
plt.figure(figsize=self.figsize)
ax = plt.subplot(1,1,1)
self.plotij(n_dimensions,i,j,ax,X_fs,X_val,X_valr,Y_val,X_test,X_testr,Y_test,ACP,index_min,index_max,classifieur_model)
plt.show()
def plot_square_piece_simple():
"""simple demo - makes more sense for a 'cut' to be the next
level of abstraction after an 'edge' though"""
pset = get_default_nub_parameters()
pset.randomize()
bxy, _ = create_puzzle_piece_edge(pset)
pset.randomize()
txy, _ = create_puzzle_piece_edge(pset)
pset.randomize()
lxy, _ = create_puzzle_piece_edge(pset)
pset.randomize()
rxy, _ = create_puzzle_piece_edge(pset)
fig = plt.figure()
fig.add_subplot(111, aspect='equal')
plt.plot(bxy[:, 0], random_sign() * bxy[:, 1], 'k-')
plt.plot(txy[:, 0], random_sign() * txy[:, 1] + 1, 'k-')
plt.plot(random_sign() * lxy[:, 1], lxy[:, 0], 'k-')
plt.plot(random_sign() * rxy[:, 1] + 1, rxy[:, 0], 'k-')
plt.show()
def getPredictedPriceNormalized(self):
# get model first
self.getModelFromFilePath(self, self.file)
input_features = self.df.iloc[:, [2, 3]].values
input_data = input_features
predicted_value = self.model.predict(self.X_test)
plt.figure(figsize=(100, 40))
plt.plot(predicted_value, color='red')
plt.plot(input_data[self.lookback:self.test_size + (2 * self.lookback),
1],
color='green')
plt.title("Opening price of stocks sold")
plt.xlabel("Time (latest-> oldest)")
plt.ylabel("Stock Opening Price")
plt.show()
self.sc.inverse_transform(input_features[self.lookback:self.test_size +
(2 * self.lookback)])
return predicted_value
def display_XY(self,n_dimensions,X_fs,X_val,Y_val,X_test,Y_test,classifieur_model):
ACP = PCA(n_components=n_dimensions)
ACP.fit(X_fs)
X_valr = ACP.tranform(X_val)
X_testr = ACP.transform(X_test)
index_min = [min(X_valr[:,i].min(),X_testr[:,i].min()) - .5 for i in range(n_dimensions)]
index_max = [max(X_valr[:,i].max(),X_testr[:,i].max()) + .5 for i in range(n_dimensions)]
plt.figure(figsize=self.figsize)
for i in range(n_dimensions - 1):
for j in range(i + 1,n_dimensions):
ax = plt.subplot(n_dimensions - 1,n_dimensions - 1,i + 1 + ((n_dimensions - 1) * (j - 1)))
self.plotij(n_dimensions,i,j,ax,X_fs,X_val,Y_val,X_test,Y_test,ACP,index_min,index_max,classifieur_model)
plt.tight_layout()
plt.show()
# Now we predict on X rather than X_test
allPredicts = []
models = [joblib.load(m) for m in modelFiles]
for i, (m, modelName) in enumerate(zip(models, modelNames)):
print i,
pred = m.predict_proba(X)[:, 1]
allPredicts.append(pred)
allPredicts = np.array(allPredicts).T
meanPreds = allPredicts.mean(axis=1)
temp.append(ROC(y, meanPreds))
totalS.append( temp )
totalS = np.array( totalS )
plt.imgshow(totalS)
plt.show()
# tmStr = dt.now().strftime('%Y-%m-%d-%H-%M-%S')
# createSubmission(meanPreds, fileName='predictions/mean_%s.csv'%tmStr)
print 'done'
import csv
open_file = open("sitka_weather_07-2018_simple.csv", "r")
csv_file = csv.reader(open_file, delimiter=",")
header_row = next(csv_file)
'''
print(header_row)
for index, column_header in enumerate(header_row):
print(index,column_header)
'''
highs = []
for row in csv_file:
highs.append(int(row[5]))
print(highs)
import matplotlib.pyploy as plt
plt.plot(highs, c="red")
plt.title("Daily High Temp, July 2018", fontsize=16)
plt.xlabel("")
plt.ylabel("Temperature (F)", fontsize=16)
plt.tick_params(axis="both", which="major", labelsize=16)
plt.show()
