def load_model():
app_settings = read_config()
model_path = app_settings['model_path']
model_file_name = app_settings['model_file']
model_file_path = path.join(model_path, model_file_name)
model = pickle.load(open(model_file_path, 'rb'))
return model
def load_data(filename, skus, country, category):
app_settings = read_config()
data_path = app_settings['data_path']
data_file_path = path.join(data_path, filename)
df = pd.read_csv(data_file_path)
df.rename(columns={
'Sku': 'sku',
'Sales': 'actualVolume',
'Week': 'forecastWeek',
'Retailer': 'accountPlanningGroupCode',
'Market': 'market',
'Category': 'category'
},
inplace=True)
cols = [
'sku', 'actualVolume', 'forecastWeek', 'accountPlanningGroupCode',
'market', 'category'
]
df = df[df['market'] == country]
df = df[df['category'] == category]
df = df[df['sku'].isin(skus)]
df = df[cols]
df_sku_sales = df.groupby(['sku', 'forecastWeek'],
as_index=False)['actualVolume'].sum()
df_sku_sales['category'] = category
df_sku_sales['market'] = country
return df_sku_sales, df
def train():
"""
Trains ARIMA model as per Region, State $ sku
& selects the best model and saves it
"""
app_settings = read_config()
data_path = app_settings['data_path']
file = app_settings['file']
file_path = path.join(data_path, file)
print(file_path)
#Load the data set
df = load_data(file_path)
#Extract the unique codes for filtering the over all data
sku_group = df.groupby('Product_SKU', as_index=False)
sku_list = sku_group.groups.keys()
region_group = df.groupby('Region', as_index=False)
region_list = region_group.groups.keys()
state_group = df.groupby('State', as_index=False)
state_list = state_group.groups.keys()
#Loops for getting the conditional data and fitting the model
for Region in region_list:
for state in state_list:
for sku in sku_list:
print('Current filtering condition :', Region, state, sku)
# load the conditional data for each condition
conditional_df = load_conditional_data(Region, state, sku, df)
data = conditional_df.Sales.reset_index(drop=True)
train, test = data[:330], data[330:365]
# #################Fit model with some validation (cv) samples ##############
arima = evaluate_models(train)
# Save model to disk
result_path = app_settings['result_path']
model_file_name = Region + state + sku + ".Pickle"
model_file_path = path.join(result_path, model_file_name)
save_model_to_disk(arima, model_file_path)
# #################saving the model as pickel file ##############
print(
'____________________________________________________________________________________________'
)
print('Training completed')
def train(filename):
"""
Trains ARIMA model post least MSE per sku & selects the best model and saves it
:return: None
"""
app_settings = read_config()
data_path = app_settings['data_path']
file_path = path.join(data_path, filename)
print(file_path)
begin = 0
end = 1
df = pd.read_excel(file_path)
#df.set_index('Order_date', inplace=True)
# ExcelFile
#df = pd.read_excel(file_path)
#df = pd.read_excel(file_path)
# Columns: Sku, Week, Sales
sku_group = df.groupby('Product SKU', as_index=False)
sku_list = sku_group.groups.keys()
region_group = df.groupby('Region', as_index=False)
region_list = region_group.groups.keys()
state_group = df.groupby('State', as_index=False)
state_list = state_group.groups.keys()
sku_best_model = []
for Region in region_list:
for state in state_list:
for sku in sku_list:
print()
print(sku)
# Select SKU to train & validate model
df_sku = df[df['Product SKU'].isin([sku])
& df['State'].isin([state])
& df['Region'].isin([Region])]
period_index = 0
best_period_models = []
for tp in train_period:
print()
# print('Begin:%d End:%d' % (tp[0], tp[1]))
print()
# Select SKU data from beginning to end of train period
df_train_period = df_sku[
(df_sku['Order_date'] >= tp[begin])
& (df_sku['Order_date'] <= tp[end])]
# Select SKU data from beginning to end of in-time validation period
df_validation_period = df_sku[(
df_sku['Order_date'] >= validation_period[period_index]
[begin]) & (df_sku['Order_date'] <=
validation_period[period_index][end])]
df_mse_period = df_sku[(
df_sku['Order_date'] >= mse_period[period_index][begin]
) & (
df_sku['Order_date'] <= mse_period[period_index][end])]
print('%d train samples for %d period.' %
(len(df_train_period), (period_index + 1)))
print('%d validation samples for %d period.' %
(len(df_validation_period), (period_index + 1)))
print('%d mse samples for %d period.' %
(len(df_mse_period), (period_index + 1)))
# Select sales data for training & validation
train_sales = df_train_period['Sales'].reset_index(
drop=True)
validation_sales = df_validation_period[
'Sales'].reset_index(drop=True)
mse_sales = df_mse_period['Sales'].reset_index(drop=True)
train_valid_set = (train_sales, validation_sales,
mse_sales)
# Evaluate best model of selected train period
#Added for my reference
print('Reference:', p_range, d_range, q_range)
#Above line
best_score, best_cfg, best_params, best_residuals, best_p, best_q, best_k_exog, best_k_trend, best_intercept, y_predict_log = evaluate_models(
train_valid_set, p_range, d_range, q_range)
best_period_model = {
'best_cfg': best_cfg,
'mse': best_score,
'sku': sku,
'week': (period_index + 1),
'residuals': best_residuals,
'p': best_p,
'q': best_q,
'k_exog': best_k_exog,
'k_trend': best_k_trend,
'State': state,
'Region': Region,
'params': best_params,
'intercept': best_intercept
}
best_period_models.append(best_period_model)
period_index += 1
# Select best model in entire perio
best_model = find_best_model(best_period_models)
# Add to best models list
sku_best_model.append(best_model)
print(
'____________________________________________________________________________________________'
)
print(
'____________________________________________________________________________________________'
)
# Save model to disk
model_path = app_settings['model_path']
file_parts = filename.split('.')
# model_file_name = file_parts[0] + '_HyperParameters.pickle'
model_file_name = 'demand_forecast.pickle'
model_file_path = path.join(model_path, model_file_name)
save_model_to_disk(model_file_path, sku_best_model)
print('Training completed')
Created on 3 Mar 2018
Project: forecastlibrary
File: forecast_service
Author: prasenjit.giri
Copyright: Accenture AI
"""
from flask import Flask, request, jsonify
from flask_cors import CORS, cross_origin
from forecast import forecast_oot
from app_settings import read_config
from forecast_service_helper import parse_json
app = Flask(__name__)
CORS(app)
@app.route('/df/v1/forecast', methods=['POST'])
@cross_origin()
def forecast():
json_string = request.get_json()
req_dict = parse_json(json_string)
result = forecast_oot(req_dict)
return jsonify(result)
if __name__ == '__main__':
app_settings = read_config()
port = int(app_settings['port'])
app.run(port=port, debug=True)