def form_to_data(form):
building_volume = form['building_volume'].data
from_date = form["from_date"].data
to_date = form['to_date'].data
building_type = form["building_type"].data
set_temp = form["set_temp"].data
rad_norm = form["rad_norm"].data
rad_hor = form["rad_hor"].data
out_temp = form["out_temp"].data
location = form['location'].data
humidity = form['humidity'].data
#use date.timetuple().tm_yday if combine both month and day to a single number
result = []
days = get_between_dates(from_date, to_date)
for day in days:
for hour in range(24):
result.append([
building_volume, day.month, day.day, hour,
getRValue(building_type), set_temp, rad_norm, rad_hor,
out_temp, humidity
])
result = np.array(result)
return result
def comparative_form_to_data(form):
building_volume = form['building_volume'].data
building_volume2 = form['building_volume2'].data
from_date = form["from_date"].data
to_date = form['to_date'].data
building_type = form["building_type"].data
building_type2 = form['building_type2'].data
set_temp = form["set_temp"].data
set_temp2 = form['set_temp2'].data
rad_norm = form["rad_norm"].data
rad_hor = form["rad_hor"].data
out_temp = form["out_temp"].data
location = form['location'].data
humidity = form['humidity'].data
humidity2 = form['humidity2'].data
result1 = []
result2 = []
days = get_between_dates(from_date, to_date)
for day in days:
for hour in range(24):
result1.append([
building_volume, day.month, day.day, hour,
getRValue(building_type), set_temp, rad_norm, rad_hor,
out_temp, humidity
])
result2.append([
building_volume2, day.month, day.day, hour,
getRValue(building_type2), set_temp2, rad_norm, rad_hor,
out_temp, humidity2
])
result1 = np.array(result1)
result2 = np.array(result2)
return result1, result2
def add_to_database(data_file):
query = '''
INSERT INTO dbo.building_data
([buildingName]
,[buildingVolume]
,[inDatetime]
,[RValue]
,[setPoint]
,[irradiationNorm]
,[irradiationHor]
,[outsideTemp]
,[location]
,[humidity]
,[heating]
,[cooling])
VALUES
(?
,?
,?
,?
,?
,?
,?
,?
,?
,?
,?
,?)'''
data = pd.read_csv(data_file, header=0)
complete_data(data)
#it seems that cnxn can't take numpy.int64 type. So converting numbers to float
vals = [(data[data.columns[0]][i],
float(data[data.columns[1]][i]),
datetime.datetime(1970, int(data[data.columns[3]][i]), int(data[data.columns[4]][i]), get_hour(int(data[data.columns[5]][i])), 0, 0),
getRValue(data[data.columns[2]][i]),
float(data[data.columns[6]][i]),
float(data[data.columns[7]][i]),
float(data[data.columns[8]][i]),
float(data[data.columns[9]][i]),
data[data.columns[10]][i],
float(data[data.columns[11]][i]),
float(data[data.columns[12]][i]),
float(data[data.columns[13]][i])) for i in range(len(data))]
cursor.executemany(query, vals)
cnxn.commit()
def file_to_data(df):
#get R value from the building type
df['RValue'] = df.apply(lambda x: getRValue(x[df.columns[2]]), axis=1)
df = df.filter(items=[
df.columns[1], df.columns[3], df.columns[4], df.columns[5],
df.columns[6], 'RValue', df.columns[7], df.columns[8], df.columns[9],
df.columns[11]
])
#fill in the data that are missing
fill_empty_data(df)
#it's possible that the entire column is missing. Replace it with 0 in that case
df.fillna(0, inplace=True)
return df.values
from sklearn.ensemble import RandomForestRegressor
from sklearn.model_selection import train_test_split
from sklearn.utils import shuffle
from sklearn.externals import joblib
import pandas as pd
from data_utils.data_utils import getRValue
if __name__ == '__main__':
df = pd.read_csv('ML_Fullstack/example_input/fake_data_2.csv', header=0)
df = shuffle(df)
df['RValue'] = df.apply(lambda x: getRValue(x[df.columns[2]]), axis=1)
data_indices = [1, 3, 4, 5, 6, 7, 8, 9, 11, 14]
label_indices = [12, 13]
data_data = df.filter(items = [df.columns[idx] for idx in data_indices]).values
data_label = df.filter(items = [df.columns[idx] for idx in label_indices]).values
classifier = RandomForestRegressor(n_estimators=100)
data_train, data_test, label_train, label_test = train_test_split(data_data, data_label, test_size=0.2,
random_state=1)
classifier.fit(data_train, label_train)
import time
start = time.time()
score = classifier.score(data_test, label_test)
end = time.time()