def save_json(json_data: str, filename: str):
conn_string = get_setting("AZURE_STORAGE_ACCOUNT")
container = get_setting("BLOB_CONTAINER_NAME")
blob_service_client = BlobServiceClient.from_connection_string(conn_string)
blob_client = blob_service_client.get_blob_client(container=container,
blob=filename)
blob_client.upload_blob(json_data)
def read_DB_table(tableName):
conn = pyodbc.connect('Driver={ODBC Driver 17 for SQL Server};'
'Server=' + get_setting("SERVER") + ';'
'Database=' + get_setting("DATABASE") + ';'
'Uid=' + get_setting("DB_USER") + ';'
'Pwd=' + get_setting("DB_PWD") + ';'
'Connection Timeout=30')
cursor = conn.cursor()
df_database = pd.read_sql_query('SELECT * FROM [dbo].[' + tableName + ']',
conn)
return df_database
def side_menu():
table_name = st.sidebar.text_input("Inserire il nome della tabella su cui fare la predizione")
if(st.sidebar.button('Esegui Predizione')):
with st.sidebar:
try:
if table_name == "":
st.error("Errore: nome tabella non valido")
else:
df = read_data.read_DB_table(table_name)
###Senza trigger###
# df_transform = preprocessing.dataset_transform(df)
# df_preprocess = preprocessing.preprocessing(df_transform)
# prediction = predict_model.score_model(df_preprocess)
# df['Prediction'] = prediction
# json_result = df.to_json(orient='split')
# utc_timestamp = datetime.datetime.utcnow().strftime("%Y%m%d%H%M%S")
# json_filename = f"prediction_{utc_timestamp}.json"
# save_prediction.save_json(json_result, json_filename)
###Con trigger###
utc_timestamp = datetime.datetime.utcnow().strftime("%Y%m%d%H%M%S")
json_filename = f"prediction_{utc_timestamp}.json"
response = requests.get(get_setting("TRIGGER_URL"), params={'table_name': table_name, 'json_filename': json_filename})
data = response.json().get("data")
columns = response.json().get("columns")
df = pd.DataFrame.from_dict(data)
df.columns = columns
outliers = df['Prediction'].loc[df.Prediction == -1].count()/df['Prediction'].count() * 100
st.write('La predizione effettuata sui dati, ha rilevato la presenza di outlier nella misura del %1.1f%%.' % outliers)
labels = 'Outliers', 'Inliers'
sizes = df.groupby('Prediction').count().Value
if sizes.count() == 1:
sizes[0] = 0
explode = (0.1, 0)
fig, ax = plt.subplots(figsize=(15,10))
ax.pie(sizes, explode=explode, labels=labels, autopct='%1.1f%%', startangle=90, textprops={'fontsize': 25})
ax.axis('equal')
st.pyplot(fig)
st.markdown(f'Per maggiori dettagli, è stato salvato nello storage il file *{json_filename}*')
except Exception as e:
st.error(f'Errore durante la predizione. \n{e}')
def score_model(data: pd.DataFrame):
model_url = get_setting("MODEL_URL")
headers = {'Authorization': f'Bearer {get_setting("DATABRICKS_TOKEN")}'}
data_json = data.to_dict(orient='split')
response = requests.request(method='POST',
headers=headers,
url=model_url,
json=data_json)
if response.status_code != 200:
raise Exception(
f'Request failed with status {response.status_code}, {response.text}'
)
return response.json()