counterweight/drogon
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

Nuevos modulos para analisis

Browse source
This commit is contained in:
pablomartincalvo 2018-12-21 19:17:39 +01:00
parent e304069684
commit d71b69a611
4 changed files with 137 additions and 1 deletions
Download patch file Download diff file Expand all files Collapse all files

37
analysis/index_batch.py Normal file
View file

@ -0,0 +1,37 @@
from analysis.market_snapshot import Market, available_date_ranges
class IndexMM:
def __init__(self):
self.name = 'indexmm'
self.date = None
self.data = None
def calculate(self, market):
self.market = market
self.date = self.market.end_date
self.data = market.get_market_data()
data_coche_pequeno = {'count': self.data[self.data['tamano_categorico'] == 'coche pequeño'].count(),
'mean': self.data[self.data['tamano_categorico' == 'coche pequeño']]['precio'].transform('mean')}
data_coche_grande = {'count': self.data[self.data['tamano_categorico'] == 'coche grande'].count(),
'mean': self.data[self.data['tamano_categorico' == 'coche grande']]['precio'].transform('mean')}
data_coche_moto = {'count': self.data[self.data['tamano_categorico'] == 'coche y moto'].count(),
'mean': self.data[self.data['tamano_categorico' == 'coche y moto']]['precio'].transform('mean')}
self.value = (((data_coche_grande['count'] * data_coche_grande['mean']) + (data_coche_moto['count'] * data_coche_moto['mean'])
+ (data_coche_pequeno['count'] * data_coche_pequeno['mean']))
/ (data_coche_grande['count'] + data_coche_moto['count'] + data_coche_pequeno['count']))
#SEGUIR AQUI
def get_data(self):
return {'name': self.name,
'date': self.date,
'value': self.value}

74
analysis/market_snapshot.py Normal file
View file

@ -0,0 +1,74 @@
from datetime import datetime, timedelta
import pandas as pd
available_date_ranges = [{'start': datetime(2017, 10, 1), 'end': datetime(2018, 1, 1)},
{'start': datetime(2017, 11, 1), 'end': datetime(2018, 2, 1)},
{'start': datetime(2017, 12, 1), 'end': datetime(2018, 3, 1)},
{'start': datetime(2018, 1, 1), 'end': datetime(2018, 4, 1)},
{'start': datetime(2018, 2, 1), 'end': datetime(2018, 5, 1)},
{'start': datetime(2018, 3, 1), 'end': datetime(2018, 6, 1)},
{'start': datetime(2018, 4, 1), 'end': datetime(2018, 7, 1)},
{'start': datetime(2018, 5, 1), 'end': datetime(2018, 8, 1)},
{'start': datetime(2018, 6, 1), 'end': datetime(2018, 9, 1)},
{'start': datetime(2018, 7, 1), 'end': datetime(2018, 10, 1)},
{'start': datetime(2018, 8, 1), 'end': datetime(2018, 11, 1)},
{'start': datetime(2018, 9, 1), 'end': datetime(2018, 12, 1)}]
class Market:
def __init__(self):
self.start_date = datetime.today() - timedelta(days=90)
self.end_date = datetime.today()
self.market = None
def load_market(self, market_query_results):
self.market = pd.DataFrame(market_query_results)
def clean_market(self, method):
if method == 'index':
self.market.dropna(subset=['tamano_categorico'])
self.market = self.market[~self.market['tamano_categorico'].isin(['2 coches o más', 'moto'])]
self.market.drop_duplicates(subset=['precio', 'latitud', 'longitud'], keep='last')
self.market = self.market[self.market['tipo_anuncio'] == 1]
self.delete_outliers()
if method == 'valoracion':
self.market.dropna(subset=['tamano_categorico'])
self.market = self.market[~self.market['tamano_categorico'].isin(['2 coches o más', 'moto'])]
self.market = self.market[self.market['precision'].isin(['ROOFTOP'])]
self.market.drop_duplicates(subset=['precio', 'latitud', 'longitud'], keep='last')
self.market = self.market[self.market['tipo_anuncio'] == 1]
self.delete_outliers()
def delete_outliers(self):
outlier_combinations = [{'tipo_anuncio': 1, 'tamano_categorico': 'coche grande',
'min_precio': 1000, 'max_precio': 150000},
{'tipo_anuncio': 1, 'tamano_categorico': 'coche pequeño',
'min_precio': 1000, 'max_precio': 150000},
{'tipo_anuncio': 1, 'tamano_categorico': 'coche y moto',
'min_precio': 1000, 'max_precio': 200000},
{'tipo_anuncio': 1, 'tamano_categorico': 'moto',
'min_precio': 1000, 'max_precio': 40000},
{'tipo_anuncio': 2, 'tamano_categorico': 'coche grande',
'min_precio': 10, 'max_precio': 300},
{'tipo_anuncio': 2, 'tamano_categorico': 'coche pequeño',
'min_precio': 10, 'max_precio': 300},
{'tipo_anuncio': 2, 'tamano_categorico': 'coche y moto',
'min_precio': 10, 'max_precio': 3000},
{'tipo_anuncio': 2, 'tamano_categorico': 'moto',
'min_precio': 10, 'max_precio': 150}]
for combination in outlier_combinations:
self.market = self.market.loc[~(
(self.market['tipo_anuncio'] == combination['tipo_anuncio']) &
(self.market['tamano_categorico'] == combination['tamano_categorico']) &
((self.market['precio'] < combination['min_precio']) | (self.market['precio'] > combination['max_precio']))
)]
def get_market_data(self):
return self.market

24
db_layer/capturas_interface.py
View file

@ -96,6 +96,30 @@ class CapturasInterface:
self.anunciosdb.query(query_statement, query_parameters)
def get_market_snapshot(self, start_date, end_date):
query_statement = """
SELECT *
FROM `anuncios`.`capturas` `t1`
WHERE
(
(
`t1`.`fecha_captura` =
(
SELECT
max(`t2`.`fecha_captura`)
FROM `anuncios`.`capturas` `t2`
WHERE (`t1`.`referencia` = `t2`.`referencia`)
)
)
AND (`t1`.`fecha_captura` BETWEEN %(start_date)S AND %(end_date)S)
)
"""
query_parameters = {'start_date': start_date.strftime('%Y-%m-%d 00:00:00'),
'end_date': end_date.strftime('%Y-%m-%d 00:00:00')}
cursor_result = self.anunciosdb.query(query_statement, query_parameters, dictionary=True)
return cursor_result.fetchall()
capturas_interface = CapturasInterface()

3
requirements.txt
View file

@ -1,4 +1,5 @@
bs4
mysql-connector-python
requests==2.18.4
html5lib
html5lib
pandas==0.23.4