# plt.figure(figsize=(10,6))
# plt.plot(data_final.date,data_final.sales)
# plt.plot(data_final.date,data_final.forecast)
# plt.savefig("/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/plots/"+ str(ref) + '.png')
#writing the result in a csv
#TODO: si forecast est negative => rendre la valeur = 0
data_final.forecast[data_final.forecast < 0] = 0
# delete hh:mm:ss part from the date
data_final['date'] = data_final['date'].astype("str").str.split(' ').str[0]
#using the interpretability model
data = data_final.copy()
data = data[['date', 'PV', 'forecast']].dropna()
list_dicts = []
kpi_sku = {"indicators": "values"}
kpi_sku.update(evaluate_all(data.PV.values, data.forecast.values))
list_dicts.append(kpi_sku)
#create the dataframe of the kpi for each sku from the dict created
kpi_df = pd.DataFrame(list_dicts, columns=list_dicts[0].keys())
kpi_df[['mse', 'mape', 'smape', 'mae', 'mda']]
data_final['mape'] = kpi_df.mape
SCHEMA_PREV = [{
"description": "date",
"mode": "REQUIRED",
"name": "date",
"type": "STRING"
}, {
"description": "nombre de pagevues reel",
"mode": "NULLABLE",
slack_message = {'text': 'saving forecasts in csv ! :)'}
requests.post(url=web_hook_url, data=json.dumps(slack_message))
#using the interpretability model
data = pd.read_csv(
"/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/tables/"
+ "RES_FINAL" + ".csv")
#using only the test set part
data = data[['date', 'sales', 'forecast', 'sku']].dropna()
DAP = diag_ability_performance(data.sales.values, data.forecast.values,
data.date.values, data.sku.values)
DAP.to_csv(
"/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/tables/"
+ "ROC" + ".csv",
sep=";",
index=False)
#this csv is used in a powerbi dashboard to visualize which sku have better performed and which ones created created a huge gap
list_dicts = []
for sku in np.unique(data.sku):
data_sku = data[data.sku == sku]
kpi_sku = {"sku": sku}
kpi_sku.update(
evaluate_all(data_sku.sales.values, data_sku.forecast.values))
list_dicts.append(kpi_sku)
#create the dataframe of the kpi for each sku from the dict created
kpi_df = pd.DataFrame(list_dicts, columns=list_dicts[0].keys())
kpi_df.to_csv(
"/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/tables/"
+ "kpi_df" + ".csv",
sep=";",
index=False)
#TODO : ameliorer le plot : ajouter 3 couleur (train, test ,futur)
# plt.figure(figsize=(10,6))
# plt.plot(data_final.date,data_final.sales)
# plt.plot(data_final.date,data_final.forecast)
# plt.savefig("/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/plots/"+ str(ref) + '.png')
#writing the result in a csv
#TODO: si forecast est negative => rendre la valeur = 0
data_final.forecast[ data_final.forecast < 0 ] = 0
data_final['date'] = data_final['date'].astype("str").str.split(' ').str[0]
#using the interpretability model
data = data_final.copy()
data = data[['date', 'max_nb_comm_hour', 'forecast']].dropna()
list_dicts = []
kpi_sku = {"indicators" : "values"}
kpi_sku.update(evaluate_all(data.max_nb_comm_hour.values,data.forecast.values))
list_dicts.append(kpi_sku)
#create the dataframe of the kpi for each sku from the dict created
kpi_df = pd.DataFrame(list_dicts,columns = list_dicts[0].keys())
kpi_df[['mse','mape','smape','mae','mda']]
data_final['mape'] = kpi_df.mape
data_final.assign(mape='kpi_df.mape')
SCHEMA_PREV = [
{
"description": "date",
"mode": "REQUIRED",
"name": "date",
"type": "STRING"
},
#TODO : ameliorer le plot : ajouter 3 couleur (train, test ,futur)
# plt.figure(figsize=(10,6))
# plt.plot(data_final.date,data_final.sales)
# plt.plot(data_final.date,data_final.forecast)
# plt.savefig("/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/plots/"+ str(ref) + '.png')
#writing the result in a csv
#TODO: si forecast est negative => rendre la valeur = 0
data_final.forecast[data_final.forecast < 0] = 0
data_final['date'] = data_final['date'].astype("str").str.split(' ').str[0]
#using the interpretability model
data = data_final.copy()
data = data[['date', 'nb_ajout_panier', 'forecast']].dropna()
list_dicts = []
kpi_sku = {"indicators": "values"}
kpi_sku.update(evaluate_all(data.nb_ajout_panier.values, data.forecast.values))
list_dicts.append(kpi_sku)
#create the dataframe of the kpi for each sku from the dict created
kpi_df = pd.DataFrame(list_dicts, columns=list_dicts[0].keys())
kpi_df[['mse', 'mape', 'smape', 'mae', 'mda']]
data_final['mape'] = kpi_df.mape
SCHEMA_PREV = [{
"description": "date",
"mode": "REQUIRED",
"name": "date",
"type": "STRING"
}, {
"description": "nombre de pagevues reel",
"mode": "NULLABLE",
#TODO : ameliorer le plot : ajouter 3 couleur (train, test ,futur)
# plt.figure(figsize=(10,6))
# plt.plot(data_final.date,data_final.sales)
# plt.plot(data_final.date,data_final.forecast)
# plt.savefig("/home/alaeddinez/MyProjects/LMFR-BigData--supply--Previsions/output/plots/"+ str(ref) + '.png')
#writing the result in a csv
#TODO: si forecast est negative => rendre la valeur = 0
data_final.forecast[data_final.forecast < 0] = 0
data_final['date'] = data_final['date'].astype("str").str.split(' ').str[0]
#using the interpretability model
data = data_final.copy()
data = data[['date', 'nb_comm', 'forecast']].dropna()
list_dicts = []
kpi_sku = {"indicators": "values"}
kpi_sku.update(evaluate_all(data.nb_comm.values, data.forecast.values))
list_dicts.append(kpi_sku)
#create the dataframe of the kpi for each sku from the dict created
kpi_df = pd.DataFrame(list_dicts, columns=list_dicts[0].keys())
kpi_df[['mse', 'mape', 'smape', 'mae', 'mda']]
data_final['mape'] = kpi_df.mape
data_final.assign(mape='kpi_df.mape')
SCHEMA_PREV = [{
"description": "date",
"mode": "REQUIRED",
"name": "date",
"type": "STRING"
}, {
"description": "nombre reel",