def get_last_trade_data(self, tkr, exchange, duration):
"""
tkr should be a valid ticker for the exchange
echange should be a valid exchange code
"""
stock = dict()
stock['q'] = tkr
stock['x'] = exchange
params = list()
params.append(stock)
self.logger.info("Queried data for " + tkr)
data = gfc.get_prices_data(params, duration)
return data
def Stock_name(cmpyname):
from googlefinance.client import get_price_data, get_prices_data, get_prices_time_data
params = [{'q':cmpyname}]
period = "5Y"
df = get_prices_data(params, period)
forecast_col = cmpyname+'_Close'
df.fillna(-99999,inplace=True)
forecast_out = int(math.ceil(0.09*len(df)))
df['label']=df[forecast_col].shift(-forecast_out)
X = np.array(df.drop([cmpyname+'_Close'],1))
X = X[:-forecast_out]
X_lately = X[-forecast_out:]
df.dropna(inplace=True)
Y = np.array(df['label'])
clf = LinearRegression(n_jobs=-1)
X_train, X_test, Y_train, Y_test = cross_validation.train_test_split(X, Y, test_size=0.2)
clf.fit(X_train, Y_train)
accuracy = clf.score(X_test, Y_test)
style.use('ggplot')
forecast_set = clf.predict(X_lately)
df['Forecast'] = np.nan
d = df.iloc[-1].name
last_date=datetime.combine(d, datetime.min.time())
last_unix = last_date.timestamp()
one_day = 86400
next_unix = last_unix + one_day
for i in forecast_set:
next_date = datetime.fromtimestamp(next_unix)
next_unix += 86400
df.loc[next_date] = [np.nan for _ in range(len(df.columns)-1)]+[i]
df.reset_index(level=None,inplace=True)
df=df.rename(index=str, columns={"index": "Date"})
df['Date'] = pd.to_datetime(df['Date'])
df = df.set_index('Date')
plt.figure(figsize=(20,8))
df[cmpyname+'_Close'].plot()
df['Forecast'].plot()
plt.legend(loc=4)
plt.xlabel('Date')
plt.ylabel('Price')
df['Forecast'].plot()
df.reset_index(level=None,inplace=True)
df=df[1050:]
#print("inside")
#var2 = "new"+str(timestamp)+".png"
plt.savefig('static/images/var4')
return(df)
def example_2(self):
params = [
# Dow Jones
{
'q': ".DJI",
'x': "INDEXDJX",
},
# NYSE COMPOSITE (DJ)
{
'q': "NYA",
'x': "INDEXNYSEGIS",
},
# S&P 500
{
'q': ".INX",
'x': "INDEXSP",
}
]
period = "1Y"
# get open, high, low, close, volume data (return pandas dataframe)
df = get_prices_data(params, period)
print(df)
def fetch_nyse_index(self):
"""
Method to fetch the daily NYSE index.
: param self
: return nyse_index(pandas Dataframe)
error_val(int)
"""
error_val = -1
try:
params = [
# NYSE COMPOSITE (DJ)
{
'q': "NYA",
'x': "INDEXNYSEGIS",
}
]
period = "5d"
nyse_index = get_prices_data(params, period)
nyse_index.index.name = INDEX
return nyse_index.tail(n=1)
except Exception as e:
print(e)
return error_val
'x': "INDEXDJX",
},
# NYSE COMPOSITE (DJ)
{
'q': "NYA",
'x': "INDEXNYSEGIS",
},
# S&P 500
{
'q': ".INX",
'x': "INDEXSP",
}
]
period = "1Y"
# get open, high, low, close, volume data (return pandas dataframe)
df = get_prices_data(params, period)
print(df)
# .DJI_Open .DJI_High .DJI_Low .DJI_Close .DJI_Volume \
# 2016-07-20 18503.12 18562.53 18495.11 18559.01 85840786
# 2016-07-21 18582.70 18622.01 18555.65 18595.03 93233337
# 2016-07-22 18589.96 18590.44 18469.67 18517.23 86803016
# 2016-07-23 18524.15 18571.30 18491.59 18570.85 87706622
# 2016-07-26 18554.49 18555.69 18452.62 18493.06 76807470
# ... ... ... ... ... ...
params = [
# Dow Jones
{
'q': ".DJI",
'x': "INDEXDJX",
},
def Stock_name(cmpyname):
from googlefinance.client import get_price_data, get_prices_data, get_prices_time_data
params = [{'q': cmpyname}]
period = "5Y"
# get open, high, low, close, volume data (return pandas dataframe)
df = get_prices_data(params, period)
df.reset_index(level=None, inplace=True)
df = df.rename(index=str, columns={"index": "Date"})
df['Date'] = pd.to_datetime(df['Date'])
df = df.set_index('Date')
df['Close_diff'] = df[cmpyname + '_Close'] - df.shift()[cmpyname +
'_Close']
df['Close_diff_log'] = np.log1p(df[cmpyname + '_Close']) - np.log1p(
df.shift()[cmpyname + '_Close'])
df = df[[cmpyname + '_Close', 'Close_diff', 'Close_diff_log']]
df = df.dropna()
train = df[0:880]
test = df[880:]
len(train), len(test)
ts = train[cmpyname + '_Close'].as_matrix()
predictions = np.empty((0), dtype=np.float32)
for i in range(len(test)):
arima_3_1_0 = ARIMA(ts, order=(3, 1, 0)).fit(dist=False)
predict = arima_3_1_0.forecast()[0]
predictions = np.hstack([predictions, predict])
ts = np.hstack([ts, predict])
predictions
nans = np.zeros(len(train))
nans[:] = np.nan
orgs = pd.concat([train[cmpyname + '_Close'], test[cmpyname + '_Close']])
orgs = pd.DataFrame({
'Date': orgs.index,
'Original': orgs.as_matrix(),
'Prediction': np.hstack([nans, predictions])
})
orgs = orgs.set_index('Date')
orgs.plot(color=['blue', 'red'])
## 1 year
train = df[734:1100]
test = df[1100:]
#return len(train), len(test)
ts = train[cmpyname + '_Close'].as_matrix()
predictions = np.empty((0), dtype=np.float32)
for i in range(len(test)):
arima_3_1_0 = ARIMA(ts, order=(3, 1, 0)).fit(dist=False)
predict = arima_3_1_0.forecast()[0]
predictions = np.hstack([predictions, predict])
ts = np.hstack([ts, predict])
nans = np.zeros(len(train))
nans[:] = np.nan
orgs = pd.concat([train[cmpyname + '_Close'], test[cmpyname + '_Close']])
orgs = pd.DataFrame({
'Date': orgs.index,
'Original': orgs.as_matrix(),
'Prediction(trend)': np.hstack([nans, predictions])
})
orgs = orgs.set_index('Date')
orgs.plot(color=['blue'])
# return (plt.show())
plt.savefig('static/images/firstimage.png')
return render_template("PredictionPage.html",
imagename="static/images/firstimage.png")
def summary_metrics(compnyname):
from googlefinance.client import get_price_data, get_prices_data, get_prices_time_data
params = [{'q': compnyname}]
period = "5Y"
df = get_prices_data(params, period)
Model_name = []
Mean_squared_error = []
R2_score = []
model = LinearRegression
clf = LinearRegression(n_jobs=-1)
X = df.iloc[:, [0, 1, 2, 4]].values
Y = df.iloc[:, 3].values
X_train, X_test, Y_train, Y_test = train_test_split(X,
Y,
test_size=0.2,
random_state=np.random)
clf.fit(X_train, Y_train)
filename1 = 'Linear_Regression_model.pckl'
pickle.dump(clf, open(filename1, 'wb'))
prediction1 = clf.predict(X_test)
MSE = mean_squared_error(Y_test, prediction1)
R2 = r2_score(Y_test, prediction1)
Model_name.append(' LinearRegression')
Mean_squared_error.append(MSE)
R2_score.append(R2)
model = RandomForestRegressor
regressor = RandomForestRegressor()
X = df.iloc[:, [0, 1, 2, 4]].values
Y = df.iloc[:, 3].values
X_train, X_test, Y_train, Y_test = train_test_split(X,
Y,
test_size=0.2,
random_state=np.random)
regressor.fit(X_train, Y_train)
filename2 = 'Random_Forest_Regressor_model.pckl'
pickle.dump(regressor, open(filename2, 'wb'))
prediction2 = regressor.predict(X_test)
MSE = mean_squared_error(Y_test, prediction2)
R2 = r2_score(Y_test, prediction2)
Model_name.append('RandomForestRegressor')
Mean_squared_error.append(MSE)
R2_score.append(R2)
model = ARIMA
df = get_prices_data(params, period)
df.reset_index(level=0, inplace=True)
df = df.rename(index=str, columns={"index": "Date"})
df['Date'] = pd.to_datetime(df['Date'])
df1 = df.set_index('Date')
df2 = df1[compnyname + '_Close']
model = ARIMA(df2, order=(3, 1, 0))
model_fit = model.fit(disp=0)
filename = 'ARIMA_model.pckl'
pickle.dump(model, open(filename, 'wb'))
X = df2.values
size = int(len(X) * 0.80)
train = X[0:size]
test = X[size:len(X)]
history = [x for x in train]
predictions = list()
for t in range(len(test)):
model = ARIMA(history, order=(5, 1, 0))
model_fit = model.fit(disp=0)
output = model_fit.forecast()
yhat = output[0]
predictions.append(yhat)
obs = test[t]
history.append(obs)
('predicted=%f, expected=%f' % (yhat, obs))
MSE = mean_squared_error(test, predictions)
R2 = r2_score(test, predictions)
Model_name.append('ARIMA')
Mean_squared_error.append(MSE)
R2_score.append(R2)
summary2 = Model_name, Mean_squared_error, R2_score
describe1 = pd.DataFrame(summary2[0], columns={"Model_Name"})
describe2 = pd.DataFrame(summary2[1], columns={"Mean_squared_error"})
describe3 = pd.DataFrame(summary2[2], columns={"R2_score"})
des = describe1.merge(describe2,
left_index=True,
right_index=True,
how='inner')
des = des.merge(describe3, left_index=True, right_index=True, how='inner')
df = des.sort_values(ascending=False, by="R2_score").reset_index(drop=True)
return (df)