import pandas as pd
import numpy as np 
import seaborn as sns
import matplotlib as mpl
import matplotlib.pyplot as plt
import plotly.express as px
from sklearn import preprocessing
from sklearn.linear_model import LinearRegression

import random
from sklearn import datasets, linear_model
from sklearn.metrics import mean_absolute_error, mean_squared_error, r2_score
from sklearn.model_selection import cross_val_score, train_test_split, KFold
from sklearn.preprocessing import StandardScaler

%matplotlib inline

Introducción

Importamos el .csv resultante del TP1

data_location = "CABA_properatti_amenities.csv"
proper_CABA = pd.read_csv(data_location)
proper_CABA.head(10)

	Unnamed: 0	property_type	place_name	state_name	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	salon	laundry	pileta	cochera	estado_nuevo	estado_a_reciclar
0	0	PH	Mataderos	Capital Federal	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	0	1	0	0	1	0	0	1	0
1	2	apartment	Mataderos	Capital Federal	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	0	0	0	0	1	0	0	0	0
2	7	apartment	Belgrano	Capital Federal	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	0	0	0	0	1	1	0	1	0
3	8	apartment	Belgrano	Capital Federal	-34.559873	-58.443362	195000.0	65.0	60.0	3000.000000	3.0	0	0	0	0	1	1	0	1	0
4	13	apartment	Palermo Soho	Capital Federal	-34.580504	-58.405874	111700.0	50.0	30.0	2234.000000	1.0	1	0	0	0	0	1	0	0	0
5	14	apartment	Palermo Soho	Capital Federal	-34.580504	-58.405874	147900.0	42.0	31.0	3521.428571	1.0	0	0	0	0	0	1	0	0	0
6	16	PH	Mataderos	Capital Federal	-34.652356	-58.501624	239000.0	140.0	98.0	1707.142857	4.0	0	0	1	0	0	0	0	0	0
7	19	apartment	Palermo	Capital Federal	-34.580504	-58.405874	350000.0	104.0	96.0	3365.384615	3.0	1	0	0	0	0	0	0	1	0
8	21	apartment	Palermo	Capital Federal	-34.590926	-58.411665	270500.0	118.0	73.0	2292.372881	4.0	0	1	0	0	0	0	0	0	0
9	29	apartment	Flores	Capital Federal	-34.635118	-58.473964	75000.0	43.0	43.0	1744.186047	2.0	0	0	0	0	0	0	0	0	0

Borramos las columnas que no nos van a servir para el analisis porque tienen datos irrelevantes

columnas_borrar=["Unnamed: 0"]
CABA = proper_CABA.drop(columnas_borrar, axis=1)
CABA.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 28762 entries, 0 to 28761
Data columns (total 19 columns):
 #   Column                 Non-Null Count  Dtype  
---  ------                 --------------  -----  
 0   property_type          28762 non-null  object 
 1   place_name             28762 non-null  object 
 2   state_name             28762 non-null  object 
 3   lat                    28762 non-null  float64
 4   lon                    28762 non-null  float64
 5   price_aprox_usd        28762 non-null  float64
 6   surface_total_in_m2    28762 non-null  float64
 7   surface_covered_in_m2  28762 non-null  float64
 8   price_per_m2           28762 non-null  float64
 9   rooms                  28762 non-null  float64
 10  terraza                28762 non-null  int64  
 11  patio                  28762 non-null  int64  
 12  parrilla               28762 non-null  int64  
 13  salon                  28762 non-null  int64  
 14  laundry                28762 non-null  int64  
 15  pileta                 28762 non-null  int64  
 16  cochera                28762 non-null  int64  
 17  estado_nuevo           28762 non-null  int64  
 18  estado_a_reciclar      28762 non-null  int64  
dtypes: float64(7), int64(9), object(3)
memory usage: 4.2+ MB

cantidad_nulos = CABA.isnull().sum()
cantidad_nulos

property_type            0
place_name               0
state_name               0
lat                      0
lon                      0
price_aprox_usd          0
surface_total_in_m2      0
surface_covered_in_m2    0
price_per_m2             0
rooms                    0
terraza                  0
patio                    0
parrilla                 0
salon                    0
laundry                  0
pileta                   0
cochera                  0
estado_nuevo             0
estado_a_reciclar        0
dtype: int64

CABA["property_type"].value_counts()

apartment    23996
house         1858
PH            1751
store         1157
Name: property_type, dtype: int64

CABA["rooms"].value_counts()

2.0     7456
3.0     7406
1.0     6592
4.0     4950
5.0     1580
6.0      424
7.0      186
8.0       80
9.0       32
10.0      29
12.0      11
11.0       9
17.0       3
22.0       1
15.0       1
14.0       1
13.0       1
Name: rooms, dtype: int64

CABA["state_name"].value_counts()

Capital Federal    28762
Name: state_name, dtype: int64

print('Mean -> ' + str(CABA.price_aprox_usd.mean()))
print('Median -> ' + str(CABA.price_aprox_usd.median()))

Mean -> 270334.0809748974
Median -> 158000.0

Graficamos para validar que los datos sean coherentes

px.box(CABA.price_aprox_usd, orientation='h')

q1 = CABA.price_aprox_usd.quantile(0.25)
q2 = CABA.price_aprox_usd.quantile(0.5)
q3 = CABA.price_aprox_usd.quantile(0.75)
q4 = CABA.price_aprox_usd.quantile(1)

inter_range = q3 - q1
outlier_high = q3 + (1.5*inter_range)
outlier_low = q1 - (1.5*inter_range)

print('q1 -> ' + str(q1))
print('q2 -> ' + str(q2))
print('q3 -> ' + str(q3))
print('q4 -> ' + str(q4))
print('inter_range -> ' + str(inter_range))
print('upper fence -> ' + str(outlier_high))

q1 -> 105000.0
q2 -> 158000.0
q3 -> 285000.0
q4 -> 46545445.0
inter_range -> 180000.0
upper fence -> 555000.0

px.box(CABA.surface_total_in_m2, orientation='h')

Vemos que hay casos donde la superficie en m2 es cero y también siguen existiendo outliers

CABA_sup_no_out_mask = ( CABA.surface_total_in_m2 > 0) & ( CABA.surface_total_in_m2 <= 12000)
CABA = CABA[CABA_sup_no_out_mask]

px.box(CABA.surface_total_in_m2, orientation='h')

cantidad_nulos = CABA.isnull().sum()
cantidad_nulos

property_type            0
place_name               0
state_name               0
lat                      0
lon                      0
price_aprox_usd          0
surface_total_in_m2      0
surface_covered_in_m2    0
price_per_m2             0
rooms                    0
terraza                  0
patio                    0
parrilla                 0
salon                    0
laundry                  0
pileta                   0
cochera                  0
estado_nuevo             0
estado_a_reciclar        0
dtype: int64

Ahora analizamos el precio por m2

px.box(CABA.price_per_m2, orientation='h')

qp1 = CABA.price_per_m2.quantile(0.25)
qp2 = CABA.price_per_m2.quantile(0.5)
qp3 = CABA.price_per_m2.quantile(0.75)
qp4 = CABA.price_per_m2.quantile(1)


inter_range_p = qp3 - qp1
outlier_high_p = qp3 + (1.5*inter_range_p)
outlier_low_p = qp1 - (1.5*inter_range_p)
outlier_high_p

print('qp1 -> ' + str(qp1))
print('qp2 -> ' + str(qp2))
print('qp3 -> ' + str(qp3))
print('qp4 -> ' + str(qp4))
print('inter_range_p -> ' + str(inter_range_p))
print('upper fence -> ' + str(outlier_high_p))

qp1 -> 1960.5263157894735
qp2 -> 2452.830188679245
qp3 -> 3063.508403361345
qp4 -> 2600000.0
inter_range_p -> 1102.9820875718713
upper fence -> 4717.981534719152

print('Mean -> ' + str(CABA.price_per_m2.mean()))
print('Median -> ' + str(CABA.price_per_m2.median()))

Mean -> 3433.135492209381
Median -> 2452.830188679245

CABA.price_per_m2.describe()

count    2.874700e+04
mean     3.433135e+03
std      1.758346e+04
min      6.730000e+00
25%      1.960526e+03
50%      2.452830e+03
75%      3.063508e+03
max      2.600000e+06
Name: price_per_m2, dtype: float64

CABA.shape

(28747, 19)

En referencia a lo observado en el gráfico y en los estadísticos descriptivos, decidimos no tener en cuenta propiedades cuyo precio por m2 sea inferior a USD 1126 o superior a USD 5422

CABA_no_out_mask = ( CABA.price_per_m2 <=  5422.193749999995) & ( CABA.price_per_m2 >= 1126.6666666666667)

CABA_no_out = CABA[CABA_no_out_mask]

CABA_no_out.shape

(25498, 19)

print('Perdemos', len(CABA)-len(CABA_no_out),'pero consideramos que esto permitirá mejorar el modelo')

Perdemos 3249 pero consideramos que esto permitirá mejorar el modelo

CABA_no_out.price_per_m2.describe()

count    25498.000000
mean      2547.964293
std        803.538220
min       1126.666667
25%       2000.000000
50%       2435.897436
75%       2941.812016
max       5421.875000
Name: price_per_m2, dtype: float64

Vemos en el dataset anterior cuantos valores toma place_name

CABA_no_out.place_name.unique()

array(['Mataderos', 'Belgrano', 'Palermo Soho', 'Palermo', 'Flores',
       'Boedo', 'Las Cañitas', 'Balvanera', 'Caballito', 'Nuñez',
       'Almagro', 'Capital Federal', 'Colegiales', 'Barrio Norte',
       'Recoleta', 'Congreso', 'Villa Crespo', 'Chacarita',
       'Puerto Madero', 'Villa Urquiza', 'Palermo Hollywood', 'Saavedra',
       'Parque Chas', 'Barracas', 'Paternal', 'Agronomía',
       'Villa Pueyrredón', 'Coghlan', 'Parque Centenario', 'San Telmo',
       'Monserrat', 'Floresta', 'Villa Devoto', 'Boca', 'San Cristobal',
       'Abasto', 'Versalles', 'Villa del Parque', 'Monte Castro',
       'Retiro', 'Parque Patricios', 'Palermo Chico', 'Liniers',
       'Centro / Microcentro', 'Once', 'Tribunales', 'Velez Sarsfield',
       'Catalinas', 'San Nicolás', 'Parque Chacabuco',
       'Parque Avellaneda', 'Constitución', 'Palermo Viejo',
       'Villa Lugano', 'Villa Luro', 'Villa General Mitre',
       'Villa Ortuzar', 'Villa Santa Rita', 'Pompeya', 'Villa Soldati',
       'Villa Real', 'Villa Riachuelo'], dtype=object)

#Borramos las observaciones que en vez de Barrio dicen "Capital Federal"
capital_0_mask=(CABA_no_out.place_name == 'Capital Federal') 
capital_0 = CABA_no_out[capital_0_mask]
capital_0.place_name.value_counts()

Capital Federal    567
Name: place_name, dtype: int64

Corresponden a publicaciones cuyo lugar fue descripto como "Capital Federal" si hacer referencia a un barrio. Decidimos borrarlos

CABA_ok_mask = CABA_no_out.place_name != 'Capital Federal'
CABA_ok = CABA_no_out[CABA_ok_mask]
CABA_ok.shape

(24931, 19)

CABA_ok.place_name.value_counts() ## 61 barrios porque se borraron los datos que eran "Capital Federal"

Belgrano           2437
Palermo            2315
Caballito          2010
Recoleta           1336
Villa Urquiza      1322
                   ... 
Velez Sarsfield      28
Pompeya              21
Villa Soldati         5
Catalinas             3
Villa Riachuelo       3
Name: place_name, Length: 61, dtype: int64

CABA_ok.head(3)

	property_type	place_name	state_name	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	salon	laundry	pileta	cochera	estado_nuevo	estado_a_reciclar
0	PH	Mataderos	Capital Federal	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	0	1	0	0	1	0	0	1	0
1	apartment	Mataderos	Capital Federal	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	0	0	0	0	1	0	0	0	0
2	apartment	Belgrano	Capital Federal	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	0	0	0	0	1	1	0	1	0

columnas_para_correlacion = ['property_type', 'place_name', 'price_aprox_usd', 'price_per_m2', 'lat', 'lon', 'surface_total_in_m2','surface_covered_in_m2','rooms']
CABA_para_correlacion = CABA_ok [columnas_para_correlacion].copy()                                                          

CABA_para_correlacion.head(3)

	property_type	place_name	price_aprox_usd	price_per_m2	lat	lon	surface_total_in_m2	surface_covered_in_m2	rooms
0	PH	Mataderos	62000.0	1127.272727	-34.661824	-58.508839	55.0	40.0	2.0
1	apartment	Mataderos	72000.0	1309.090909	-34.652262	-58.522982	55.0	55.0	2.0
2	apartment	Belgrano	138000.0	3066.666667	-34.559873	-58.443362	45.0	40.0	1.0

# visualizamos la matriz de correlación en Seaborn usando a heatmap

sns.heatmap(CABA_para_correlacion.corr(), vmin=-1, vmax=1, center=0, cmap="YlGnBu");

DUMMIES

Generamos variables dummies para poder trabajar con variables categóricas

#Drop_first true para no tener multicolinealidad, se fue la columna de dummies PH 
dummy_prop_type = pd.get_dummies(CABA_ok['property_type'],drop_first=True)
dummy_prop_type.head()

	apartment	house	store
0	0	0	0
1	1	0	0
2	1	0	0
3	1	0	0
4	1	0	0

CABA_ok.place_name.sort_values().unique()

array(['Abasto', 'Agronomía', 'Almagro', 'Balvanera', 'Barracas',
       'Barrio Norte', 'Belgrano', 'Boca', 'Boedo', 'Caballito',
       'Catalinas', 'Centro / Microcentro', 'Chacarita', 'Coghlan',
       'Colegiales', 'Congreso', 'Constitución', 'Flores', 'Floresta',
       'Las Cañitas', 'Liniers', 'Mataderos', 'Monserrat', 'Monte Castro',
       'Nuñez', 'Once', 'Palermo', 'Palermo Chico', 'Palermo Hollywood',
       'Palermo Soho', 'Palermo Viejo', 'Parque Avellaneda',
       'Parque Centenario', 'Parque Chacabuco', 'Parque Chas',
       'Parque Patricios', 'Paternal', 'Pompeya', 'Puerto Madero',
       'Recoleta', 'Retiro', 'Saavedra', 'San Cristobal', 'San Nicolás',
       'San Telmo', 'Tribunales', 'Velez Sarsfield', 'Versalles',
       'Villa Crespo', 'Villa Devoto', 'Villa General Mitre',
       'Villa Lugano', 'Villa Luro', 'Villa Ortuzar', 'Villa Pueyrredón',
       'Villa Real', 'Villa Riachuelo', 'Villa Santa Rita',
       'Villa Soldati', 'Villa Urquiza', 'Villa del Parque'], dtype=object)

#Drop_first true para no tener multicolinealidad, se fue la columna de dummies 'Abasto'
dummy_barrio = pd.get_dummies(CABA_ok['place_name'],drop_first=True)
dummy_barrio.head()

	Agronomía	Almagro	Balvanera	Barracas	Barrio Norte	Belgrano	Boca	Boedo	Caballito	Catalinas	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
0	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
1	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
2	0	0	0	0	0	1	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
3	0	0	0	0	0	1	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
4	0	0	0	0	0	0	0	0	0	0	...	0	0	0	0	0	0	0	0	0	0

5 rows × 60 columns

# concatenamos los datasets 
CABA_con_dummies = pd.concat([CABA_ok,dummy_prop_type,dummy_barrio],axis=1)

CABA_con_dummies.head(10)

	property_type	place_name	state_name	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
0	PH	Mataderos	Capital Federal	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	...	0	0	0	0	0	0	0	0	0	0
1	apartment	Mataderos	Capital Federal	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	...	0	0	0	0	0	0	0	0	0	0
2	apartment	Belgrano	Capital Federal	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	...	0	0	0	0	0	0	0	0	0	0
3	apartment	Belgrano	Capital Federal	-34.559873	-58.443362	195000.0	65.0	60.0	3000.000000	3.0	...	0	0	0	0	0	0	0	0	0	0
4	apartment	Palermo Soho	Capital Federal	-34.580504	-58.405874	111700.0	50.0	30.0	2234.000000	1.0	...	0	0	0	0	0	0	0	0	0	0
5	apartment	Palermo Soho	Capital Federal	-34.580504	-58.405874	147900.0	42.0	31.0	3521.428571	1.0	...	0	0	0	0	0	0	0	0	0	0
6	PH	Mataderos	Capital Federal	-34.652356	-58.501624	239000.0	140.0	98.0	1707.142857	4.0	...	0	0	0	0	0	0	0	0	0	0
7	apartment	Palermo	Capital Federal	-34.580504	-58.405874	350000.0	104.0	96.0	3365.384615	3.0	...	0	0	0	0	0	0	0	0	0	0
8	apartment	Palermo	Capital Federal	-34.590926	-58.411665	270500.0	118.0	73.0	2292.372881	4.0	...	0	0	0	0	0	0	0	0	0	0
9	apartment	Flores	Capital Federal	-34.635118	-58.473964	75000.0	43.0	43.0	1744.186047	2.0	...	0	0	0	0	0	0	0	0	0	0

10 rows × 82 columns

#Borramos las columnas que no nos van a servir para la regresion porque tienen datos en formato string#
columnas_borrar=["property_type", "place_name", "state_name"]
CABA_para_regresion = CABA_con_dummies.drop(columnas_borrar, axis=1)
CABA_para_regresion.shape

(24931, 79)

CABA_para_regresion.head(4)

	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
0	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	0	1	0	...	0	0	0	0	0	0	0	0	0	0
1	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
2	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
3	-34.559873	-58.443362	195000.0	65.0	60.0	3000.000000	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0

4 rows × 79 columns

#chequeamos si hay nulos
CABA_para_regresion.isnull().values.any()

False

CABA_para_regresion.info()

<class 'pandas.core.frame.DataFrame'>
Int64Index: 24931 entries, 0 to 28760
Data columns (total 79 columns):
 #   Column                 Non-Null Count  Dtype  
---  ------                 --------------  -----  
 0   lat                    24931 non-null  float64
 1   lon                    24931 non-null  float64
 2   price_aprox_usd        24931 non-null  float64
 3   surface_total_in_m2    24931 non-null  float64
 4   surface_covered_in_m2  24931 non-null  float64
 5   price_per_m2           24931 non-null  float64
 6   rooms                  24931 non-null  float64
 7   terraza                24931 non-null  int64  
 8   patio                  24931 non-null  int64  
 9   parrilla               24931 non-null  int64  
 10  salon                  24931 non-null  int64  
 11  laundry                24931 non-null  int64  
 12  pileta                 24931 non-null  int64  
 13  cochera                24931 non-null  int64  
 14  estado_nuevo           24931 non-null  int64  
 15  estado_a_reciclar      24931 non-null  int64  
 16  apartment              24931 non-null  uint8  
 17  house                  24931 non-null  uint8  
 18  store                  24931 non-null  uint8  
 19  Agronomía              24931 non-null  uint8  
 20  Almagro                24931 non-null  uint8  
 21  Balvanera              24931 non-null  uint8  
 22  Barracas               24931 non-null  uint8  
 23  Barrio Norte           24931 non-null  uint8  
 24  Belgrano               24931 non-null  uint8  
 25  Boca                   24931 non-null  uint8  
 26  Boedo                  24931 non-null  uint8  
 27  Caballito              24931 non-null  uint8  
 28  Catalinas              24931 non-null  uint8  
 29  Centro / Microcentro   24931 non-null  uint8  
 30  Chacarita              24931 non-null  uint8  
 31  Coghlan                24931 non-null  uint8  
 32  Colegiales             24931 non-null  uint8  
 33  Congreso               24931 non-null  uint8  
 34  Constitución           24931 non-null  uint8  
 35  Flores                 24931 non-null  uint8  
 36  Floresta               24931 non-null  uint8  
 37  Las Cañitas            24931 non-null  uint8  
 38  Liniers                24931 non-null  uint8  
 39  Mataderos              24931 non-null  uint8  
 40  Monserrat              24931 non-null  uint8  
 41  Monte Castro           24931 non-null  uint8  
 42  Nuñez                  24931 non-null  uint8  
 43  Once                   24931 non-null  uint8  
 44  Palermo                24931 non-null  uint8  
 45  Palermo Chico          24931 non-null  uint8  
 46  Palermo Hollywood      24931 non-null  uint8  
 47  Palermo Soho           24931 non-null  uint8  
 48  Palermo Viejo          24931 non-null  uint8  
 49  Parque Avellaneda      24931 non-null  uint8  
 50  Parque Centenario      24931 non-null  uint8  
 51  Parque Chacabuco       24931 non-null  uint8  
 52  Parque Chas            24931 non-null  uint8  
 53  Parque Patricios       24931 non-null  uint8  
 54  Paternal               24931 non-null  uint8  
 55  Pompeya                24931 non-null  uint8  
 56  Puerto Madero          24931 non-null  uint8  
 57  Recoleta               24931 non-null  uint8  
 58  Retiro                 24931 non-null  uint8  
 59  Saavedra               24931 non-null  uint8  
 60  San Cristobal          24931 non-null  uint8  
 61  San Nicolás            24931 non-null  uint8  
 62  San Telmo              24931 non-null  uint8  
 63  Tribunales             24931 non-null  uint8  
 64  Velez Sarsfield        24931 non-null  uint8  
 65  Versalles              24931 non-null  uint8  
 66  Villa Crespo           24931 non-null  uint8  
 67  Villa Devoto           24931 non-null  uint8  
 68  Villa General Mitre    24931 non-null  uint8  
 69  Villa Lugano           24931 non-null  uint8  
 70  Villa Luro             24931 non-null  uint8  
 71  Villa Ortuzar          24931 non-null  uint8  
 72  Villa Pueyrredón       24931 non-null  uint8  
 73  Villa Real             24931 non-null  uint8  
 74  Villa Riachuelo        24931 non-null  uint8  
 75  Villa Santa Rita       24931 non-null  uint8  
 76  Villa Soldati          24931 non-null  uint8  
 77  Villa Urquiza          24931 non-null  uint8  
 78  Villa del Parque       24931 non-null  uint8  
dtypes: float64(7), int64(9), uint8(63)
memory usage: 4.7 MB

columnas = list(CABA_para_regresion.columns)
print(columnas)

['lat', 'lon', 'price_aprox_usd', 'surface_total_in_m2', 'surface_covered_in_m2', 'price_per_m2', 'rooms', 'terraza', 'patio', 'parrilla', 'salon', 'laundry', 'pileta', 'cochera', 'estado_nuevo', 'estado_a_reciclar', 'apartment', 'house', 'store', 'Agronomía', 'Almagro', 'Balvanera', 'Barracas', 'Barrio Norte', 'Belgrano', 'Boca', 'Boedo', 'Caballito', 'Catalinas', 'Centro / Microcentro', 'Chacarita', 'Coghlan', 'Colegiales', 'Congreso', 'Constitución', 'Flores', 'Floresta', 'Las Cañitas', 'Liniers', 'Mataderos', 'Monserrat', 'Monte Castro', 'Nuñez', 'Once', 'Palermo', 'Palermo Chico', 'Palermo Hollywood', 'Palermo Soho', 'Palermo Viejo', 'Parque Avellaneda', 'Parque Centenario', 'Parque Chacabuco', 'Parque Chas', 'Parque Patricios', 'Paternal', 'Pompeya', 'Puerto Madero', 'Recoleta', 'Retiro', 'Saavedra', 'San Cristobal', 'San Nicolás', 'San Telmo', 'Tribunales', 'Velez Sarsfield', 'Versalles', 'Villa Crespo', 'Villa Devoto', 'Villa General Mitre', 'Villa Lugano', 'Villa Luro', 'Villa Ortuzar', 'Villa Pueyrredón', 'Villa Real', 'Villa Riachuelo', 'Villa Santa Rita', 'Villa Soldati', 'Villa Urquiza', 'Villa del Parque']

columnas.remove('price_per_m2')
print(columnas)

['lat', 'lon', 'price_aprox_usd', 'surface_total_in_m2', 'surface_covered_in_m2', 'rooms', 'terraza', 'patio', 'parrilla', 'salon', 'laundry', 'pileta', 'cochera', 'estado_nuevo', 'estado_a_reciclar', 'apartment', 'house', 'store', 'Agronomía', 'Almagro', 'Balvanera', 'Barracas', 'Barrio Norte', 'Belgrano', 'Boca', 'Boedo', 'Caballito', 'Catalinas', 'Centro / Microcentro', 'Chacarita', 'Coghlan', 'Colegiales', 'Congreso', 'Constitución', 'Flores', 'Floresta', 'Las Cañitas', 'Liniers', 'Mataderos', 'Monserrat', 'Monte Castro', 'Nuñez', 'Once', 'Palermo', 'Palermo Chico', 'Palermo Hollywood', 'Palermo Soho', 'Palermo Viejo', 'Parque Avellaneda', 'Parque Centenario', 'Parque Chacabuco', 'Parque Chas', 'Parque Patricios', 'Paternal', 'Pompeya', 'Puerto Madero', 'Recoleta', 'Retiro', 'Saavedra', 'San Cristobal', 'San Nicolás', 'San Telmo', 'Tribunales', 'Velez Sarsfield', 'Versalles', 'Villa Crespo', 'Villa Devoto', 'Villa General Mitre', 'Villa Lugano', 'Villa Luro', 'Villa Ortuzar', 'Villa Pueyrredón', 'Villa Real', 'Villa Riachuelo', 'Villa Santa Rita', 'Villa Soldati', 'Villa Urquiza', 'Villa del Parque']

CABA_para_regresion

	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
0	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	0	1	0	...	0	0	0	0	0	0	0	0	0	0
1	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
2	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
3	-34.559873	-58.443362	195000.0	65.0	60.0	3000.000000	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
4	-34.580504	-58.405874	111700.0	50.0	30.0	2234.000000	1.0	1	0	0	...	0	0	0	0	0	0	0	0	0	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
28755	-34.559873	-58.443362	128000.0	38.0	35.0	3368.421053	3.0	1	0	1	...	0	0	0	0	0	0	0	0	0	0
28756	-34.587425	-58.397372	165000.0	44.0	39.0	3750.000000	1.0	0	0	1	...	0	0	0	0	0	0	0	0	0	0
28757	-34.563685	-58.442683	410000.0	157.0	157.0	2611.464968	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
28758	-34.563685	-58.442683	410000.0	157.0	157.0	2611.464968	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
28760	-34.570639	-58.475596	131500.0	46.0	39.0	2858.695652	1.0	1	0	1	...	0	0	0	0	0	0	0	0	1	0

24931 rows × 79 columns

CABA_100= CABA_para_regresion.sample(n=100 , random_state=123)
eliminar=CABA_100.index
CABA_para_regresion=CABA_para_regresion.drop(index=eliminar).copy()

CABA_para_regresion

	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
0	-34.661824	-58.508839	62000.0	55.0	40.0	1127.272727	2.0	0	1	0	...	0	0	0	0	0	0	0	0	0	0
1	-34.652262	-58.522982	72000.0	55.0	55.0	1309.090909	2.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
2	-34.559873	-58.443362	138000.0	45.0	40.0	3066.666667	1.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
3	-34.559873	-58.443362	195000.0	65.0	60.0	3000.000000	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
4	-34.580504	-58.405874	111700.0	50.0	30.0	2234.000000	1.0	1	0	0	...	0	0	0	0	0	0	0	0	0	0
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
28755	-34.559873	-58.443362	128000.0	38.0	35.0	3368.421053	3.0	1	0	1	...	0	0	0	0	0	0	0	0	0	0
28756	-34.587425	-58.397372	165000.0	44.0	39.0	3750.000000	1.0	0	0	1	...	0	0	0	0	0	0	0	0	0	0
28757	-34.563685	-58.442683	410000.0	157.0	157.0	2611.464968	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
28758	-34.563685	-58.442683	410000.0	157.0	157.0	2611.464968	3.0	0	0	0	...	0	0	0	0	0	0	0	0	0	0
28760	-34.570639	-58.475596	131500.0	46.0	39.0	2858.695652	1.0	1	0	1	...	0	0	0	0	0	0	0	0	1	0

24831 rows × 79 columns

Regresión Lineal con Sklearn

feature_cols = columnas

X = CABA_para_regresion[feature_cols]
y = CABA_para_regresion.price_per_m2

CABA_Xtrain, CABA_Xtest, CABA_ytrain, CABA_ytest = train_test_split(X, y, random_state=123)

# creamos nuevamente X and y
X = CABA_Xtrain
y = CABA_ytrain

# creamos el modelo y fiteamos
linreg = LinearRegression()
model = linreg.fit(X, y)

# Imprimimos coeficientes
print (linreg.intercept_)
print (linreg.coef_)

217196.3821238267
[ 1.02520984e+03  3.07249682e+03  3.19019848e-03 -8.21742636e+00
  1.02615651e+00 -4.54176250e+01 -6.09039917e+01 -1.66820254e+02
  2.61586628e+01  4.41905602e+01 -3.41471580e+01  2.24830647e+02
  1.27879843e+02  3.55994728e+01 -2.47481763e+02  5.60148062e+00
  2.20716823e+00  2.51580423e+02  2.16707724e+02  1.22687482e+02
 -6.20010777e+01 -1.19880131e+02  5.15327802e+02  6.58563829e+02
 -3.58954457e+02 -2.56029994e+02  3.04972122e+02 -1.59060623e+01
  9.83944646e+01  9.54493713e+01  3.91168483e+02  4.19238624e+02
 -9.85808386e+01 -3.73616513e+02  1.33380994e+02  1.42040682e+02
  8.03255723e+02  2.93606645e+02  2.89604845e+02 -8.47438875e+01
  2.70968335e+02  6.54761171e+02 -1.06151432e+02  6.53885306e+02
  8.58386292e+02  6.83984588e+02  7.23118762e+02  5.67378270e+02
 -2.97325902e+01  2.84407441e+02  2.68687329e+01  2.03213634e+02
 -9.34817494e+01  1.27724505e+02 -4.03142403e+01  1.38711167e+03
  6.42091597e+02  3.54576871e+02  3.12520854e+02 -2.89593334e+01
  2.07291690e+01 -9.08469342e+00 -4.20840893e+01  1.91438853e+02
  4.19774306e+02  3.80407900e+02  4.56834492e+02  1.97157105e+02
  4.24506898e+01  2.63259509e+02  5.99002679e+02  4.02594486e+02
  3.45102994e+02 -1.54090138e+01  2.13649109e+02 -2.67991359e+02
  4.48548134e+02  3.40922351e+02]

# imprimos la metrica que mide la bondad de ajusto del modelo. En este caso el R2.
print ('R2_train=', ' ', model.score(X, y)) ### R2_train= 0.5773183568859408 (CON COMUNAS)

R2_train=   0.669454735944299

# Ahora probamos con los valores de test
Xt=CABA_Xtest
yt=CABA_ytest
print ('R2_test=', ' ', model.score(Xt, yt)) #R2_test= 0.5319143306475157 (CON COMUNAS)

R2_test=   0.6695211106514514

## para observarlo mejor miramos el nombre con el coeficiente
list(zip(feature_cols, linreg.coef_))

[('lat', 1025.209837371704),
 ('lon', 3072.496819612948),
 ('price_aprox_usd', 0.0031901984768509806),
 ('surface_total_in_m2', -8.217426363745773),
 ('surface_covered_in_m2', 1.0261565080286346),
 ('rooms', -45.41762499038754),
 ('terraza', -60.903991667226194),
 ('patio', -166.82025434712338),
 ('parrilla', 26.158662753751884),
 ('salon', 44.19056018840561),
 ('laundry', -34.14715804334),
 ('pileta', 224.830646822042),
 ('cochera', 127.87984265979185),
 ('estado_nuevo', 35.59947282933214),
 ('estado_a_reciclar', -247.48176339469396),
 ('apartment', 5.601480616483286),
 ('house', 2.2071682323957162),
 ('store', 251.58042252846744),
 ('Agronomía', 216.7077235211785),
 ('Almagro', 122.68748174356624),
 ('Balvanera', -62.00107773514246),
 ('Barracas', -119.88013083964427),
 ('Barrio Norte', 515.3278021934572),
 ('Belgrano', 658.5638286456942),
 ('Boca', -358.95445739148687),
 ('Boedo', -256.02999429034344),
 ('Caballito', 304.97212223470706),
 ('Catalinas', -15.906062313626267),
 ('Centro / Microcentro', 98.39446458149374),
 ('Chacarita', 95.44937134993178),
 ('Coghlan', 391.1684833891707),
 ('Colegiales', 419.2386238645372),
 ('Congreso', -98.58083861776933),
 ('Constitución', -373.6165130290824),
 ('Flores', 133.38099381148209),
 ('Floresta', 142.04068157035982),
 ('Las Cañitas', 803.255722574287),
 ('Liniers', 293.60664477855124),
 ('Mataderos', 289.6048454025032),
 ('Monserrat', -84.74388748279918),
 ('Monte Castro', 270.96833510740885),
 ('Nuñez', 654.7611710722172),
 ('Once', -106.15143170703455),
 ('Palermo', 653.8853059388196),
 ('Palermo Chico', 858.386291634913),
 ('Palermo Hollywood', 683.9845879730714),
 ('Palermo Soho', 723.118761990414),
 ('Palermo Viejo', 567.3782697964153),
 ('Parque Avellaneda', -29.73259019853296),
 ('Parque Centenario', 284.4074405743081),
 ('Parque Chacabuco', 26.86873290552856),
 ('Parque Chas', 203.21363351067185),
 ('Parque Patricios', -93.48174944380685),
 ('Paternal', 127.72450537132985),
 ('Pompeya', -40.3142403307216),
 ('Puerto Madero', 1387.111666198594),
 ('Recoleta', 642.091597192173),
 ('Retiro', 354.57687080383016),
 ('Saavedra', 312.5208537664762),
 ('San Cristobal', -28.95933344689655),
 ('San Nicolás', 20.72916898931026),
 ('San Telmo', -9.08469342270223),
 ('Tribunales', -42.084089340918595),
 ('Velez Sarsfield', 191.43885325045528),
 ('Versalles', 419.77430649660835),
 ('Villa Crespo', 380.4078998775279),
 ('Villa Devoto', 456.8344922746144),
 ('Villa General Mitre', 197.15710496664894),
 ('Villa Lugano', 42.4506897883104),
 ('Villa Luro', 263.2595090829028),
 ('Villa Ortuzar', 599.0026792612895),
 ('Villa Pueyrredón', 402.5944855970647),
 ('Villa Real', 345.1029942050745),
 ('Villa Riachuelo', -15.409013770501906),
 ('Villa Santa Rita', 213.6491093537947),
 ('Villa Soldati', -267.9913586394433),
 ('Villa Urquiza', 448.54813404860795),
 ('Villa del Parque', 340.92235054920934)]

VISUALIZACIONES

##Vamos a normalizar las variables para graficarlas
#### Normalizamos todas las variables 

rooms_norm = CABA_para_regresion["rooms"]
rooms_min = np.min(rooms_norm)
rooms_max = np.max(rooms_norm)
rooms_norm = [(x - rooms_min) / (rooms_max - rooms_min) for x in rooms_norm]


lat_norm = CABA_para_regresion["lat"]
lat_min = np.min(lat_norm)
lat_max = np.max(lat_norm)
lat_norm = [(x - lat_min) / (lat_max - lat_min) for x in lat_norm]

lon_norm = CABA_para_regresion["lon"]
lon_min = np.min(lon_norm)
lon_max = np.max(lon_norm)
lon_norm = [(x - lon_min) / (lon_max - lon_min) for x in lon_norm]

price_aprox_usd_norm = CABA_para_regresion["price_aprox_usd"]
price_aprox_usd_min = np.min(price_aprox_usd_norm)
price_aprox_usd_max = np.max(price_aprox_usd_norm)
price_aprox_usd_norm = [(x - price_aprox_usd_min) / (price_aprox_usd_max - price_aprox_usd_min) for x in price_aprox_usd_norm]


surface_total_in_m2_norm = CABA_para_regresion["surface_total_in_m2"]
surface_total_in_m2_min = np.min(surface_total_in_m2_norm)
surface_total_in_m2_max = np.max(surface_total_in_m2_norm)
surface_total_in_m2_norm = [(x - surface_total_in_m2_min) / (surface_total_in_m2_max - surface_total_in_m2_min) for x in surface_total_in_m2_norm]


surface_covered_in_m2_norm = CABA_para_regresion["surface_covered_in_m2"]
surface_covered_in_m2_min = np.min(surface_covered_in_m2_norm)
surface_covered_in_m2_max = np.max(surface_covered_in_m2_norm)
surface_covered_in_m2_norm = [(x - surface_covered_in_m2_min) / (surface_covered_in_m2_max - surface_covered_in_m2_min) for x in surface_covered_in_m2_norm]

price_per_m2_norm = CABA_para_regresion["price_per_m2"]
price_per_m2_min = np.min(price_per_m2_norm)
price_per_m2_max = np.max(price_per_m2_norm)
price_per_m2_norm = [(x - price_per_m2_min) / (price_per_m2_max - price_per_m2_min) for x in price_per_m2_norm]

#generamos un dataframe con los valores normalizados
CABA_NORM= pd.DataFrame(list(zip(lat_norm,lon_norm,price_aprox_usd_norm,surface_total_in_m2_norm,surface_covered_in_m2_norm,price_per_m2_norm,rooms_norm)),columns =['lat', 'lon','price_aprox_usd','surface_total_in_m2','surface_covered_in_m2','price_per_m2','rooms'])

# Exploramos más features

feature_cols_norm = ["lat", "lon", "surface_total_in_m2", "price_aprox_usd", "surface_covered_in_m2", "rooms"]

# plots múltiples en seaborn

sns.pairplot(CABA_NORM, x_vars=feature_cols_norm, y_vars='price_per_m2', kind='reg');

# matriz de correlación (rangos de 1 a -1)

CABA_para_regresion.corr()

	lat	lon	price_aprox_usd	surface_total_in_m2	surface_covered_in_m2	price_per_m2	rooms	terraza	patio	parrilla	...	Villa Lugano	Villa Luro	Villa Ortuzar	Villa Pueyrredón	Villa Real	Villa Riachuelo	Villa Santa Rita	Villa Soldati	Villa Urquiza	Villa del Parque
lat	1.000000	-0.250400	0.123326	0.044971	0.036446	0.258045	0.028010	0.042470	-0.032329	0.020605	...	-0.112299	-0.119924	0.025149	0.030233	-0.018255	-0.026251	-0.017620	-0.022443	0.167541	-0.029801
lon	-0.250400	1.000000	0.084947	0.015080	0.030348	0.153849	-0.015349	-0.138851	-0.117395	-0.098623	...	-0.064157	-0.205648	-0.047801	-0.139124	-0.076916	-0.009368	-0.047923	-0.008279	-0.298199	-0.163599
price_aprox_usd	0.123326	0.084947	1.000000	0.860070	0.819020	0.312747	0.469777	0.107696	0.065623	0.014915	...	-0.022108	-0.031635	0.029713	-0.014195	-0.005928	-0.003451	-0.008605	-0.007258	-0.050904	-0.026118
surface_total_in_m2	0.044971	0.015080	0.860070	1.000000	0.937360	-0.058111	0.494506	0.131901	0.165543	0.023312	...	0.000861	-0.013642	0.055917	0.002628	0.007758	0.002570	0.005335	-0.001579	-0.042118	-0.008089
surface_covered_in_m2	0.036446	0.030348	0.819020	0.937360	1.000000	-0.029964	0.456769	0.086894	0.127717	-0.003818	...	0.003908	-0.014553	0.057019	-0.001780	0.003973	0.001454	-0.003259	-0.000322	-0.042593	-0.010052
...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...	...
Villa Riachuelo	-0.026251	-0.009368	-0.003451	0.002570	0.001454	-0.016824	0.005810	0.010058	0.006422	-0.006653	...	-0.000670	-0.001322	-0.000617	-0.000880	-0.000376	1.000000	-0.000474	-0.000156	-0.002600	-0.001437
Villa Santa Rita	-0.017620	-0.047923	-0.008605	0.005335	-0.003259	-0.030268	-0.001865	0.008242	0.032497	0.003522	...	-0.002627	-0.005181	-0.002418	-0.003447	-0.001473	-0.000474	1.000000	-0.000611	-0.010192	-0.005630
Villa Soldati	-0.022443	-0.008279	-0.007258	-0.001579	-0.000322	-0.022017	0.008205	0.004396	0.011067	-0.008590	...	-0.000865	-0.001707	-0.000797	-0.001136	-0.000485	-0.000156	-0.000611	1.000000	-0.003357	-0.001855
Villa Urquiza	0.167541	-0.298199	-0.050904	-0.042118	-0.042593	-0.016778	-0.029891	0.082089	0.014456	0.090954	...	-0.014426	-0.028450	-0.013280	-0.018931	-0.008089	-0.002600	-0.010192	-0.003357	1.000000	-0.030917
Villa del Parque	-0.029801	-0.163599	-0.026118	-0.008089	-0.010052	-0.050128	0.039210	-0.011309	0.022108	-0.017554	...	-0.007970	-0.015717	-0.007336	-0.010458	-0.004469	-0.001437	-0.005630	-0.001855	-0.030917	1.000000

79 rows × 79 columns

# visualizamos la matriz de correlación en Seaborn usando a heatmap

sns.heatmap(CABA_para_regresion.corr(), vmin=-1, vmax=1, center=0, cmap="YlGnBu");

Regresión con LASSO CV

cv = KFold(10, shuffle=True)

lassocv = linear_model.LassoCV(normalize=True).fit(CABA_Xtrain, CABA_ytrain)
scores = cross_val_score(lassocv, CABA_Xtrain, CABA_ytrain, cv=cv, scoring='r2')
print ("r^2:", lassocv.score(CABA_Xtrain, CABA_ytrain))
print ("alpha:", lassocv.alpha_)

r^2: 0.6691990257988953
alpha: 0.003199590044391265

scores = cross_val_score(lassocv, CABA_Xtest, CABA_ytest, cv=cv, scoring='r2')
print ("r^2:", lassocv.score(CABA_Xtest, CABA_ytest))
print ("alpha:", lassocv.alpha_)

r^2: 0.6694694617747674
alpha: 0.003199590044391265

Regresión con LASSO

lasso = linear_model.Lasso(alpha=0.002811914108344165, normalize=True)

# Ajustamos nuevamente, esta vez con regularizacion
X = CABA_Xtrain
y = CABA_ytrain
lasso_model =lasso.fit(X, y)
predictions = lasso_model.predict(X)

print ("r^2:", lasso_model.score(X, y)) #con alpha = 0.5 --> r^2: 0.5098385825168807 (con comuna)  y 0.4904763180070667 (sin comuna)

r^2: 0.6692369809467804

Xt=CABA_Xtest
yt=CABA_ytest
print ("r^2:", lasso_model.score(Xt, yt)) #con alpha = 0.5 --> r^2: 0.4875681932460285 (con comuna)  y  0.5092279239407673 (sin comuna)

r^2: 0.6694835898509823

Regresión con RIDGE

rlm = linear_model.Ridge(alpha=0.0001, normalize=True)

# Ajustamos nuevamente, esta vez con regularizacion
X = CABA_Xtrain
y = CABA_ytrain

ridge_model = rlm.fit(X, y)
predictions = ridge_model.predict(X)

print ("r^2:", ridge_model.score(X, y)) #con alpha = 0.5 --> r^2: 0.496124980682607 (con comuna) y 0.4893773534731626 (sin comuna)

r^2: 0.6694540257063422

Xt=CABA_Xtest
yt=CABA_ytest
print ("r^2:", ridge_model.score(Xt, yt)) #con alpha = 0.5 --> r^2: 0.47811073280496075 (con comuna) y 0.4729377985558373 (sin comuna)

r^2: 0.6694969474454349

Regresión RIDGE con CV

# instanciamos un modelo 
rlmcv = linear_model.RidgeCV(alphas=np.linspace(0.0001,0.01, 1000), cv=5, normalize=True,scoring='r2')


# Ajustamos nuevamente nuestro modelo, esta vez con RidgeCV
X = CABA_Xtrain
y = CABA_ytrain
rlmcv.fit(X, y)
predictions = rlmcv.predict(X)
print ("r^2:", rlmcv.score(X, y)) # r^2: 0.5592333948827667 (CON COMUNA)

r^2: 0.6678952031855221

rlmcv.alpha_

0.007591891891891892

Xt=CABA_Xtest
yt=CABA_ytest
print ("r^2:", rlmcv.score(Xt, yt)) # r^2: 0.5240608432715783 (CON COMUNA)

r^2: 0.6661970535458268

print ("alpha:", rlmcv.alphas)

alpha: [0.0001     0.00010991 0.00011982 0.00012973 0.00013964 0.00014955
 0.00015946 0.00016937 0.00017928 0.00018919 0.0001991  0.00020901
 0.00021892 0.00022883 0.00023874 0.00024865 0.00025856 0.00026847
 0.00027838 0.00028829 0.0002982  0.00030811 0.00031802 0.00032793
 0.00033784 0.00034775 0.00035766 0.00036757 0.00037748 0.00038739
 0.0003973  0.00040721 0.00041712 0.00042703 0.00043694 0.00044685
 0.00045676 0.00046667 0.00047658 0.00048649 0.0004964  0.00050631
 0.00051622 0.00052613 0.00053604 0.00054595 0.00055586 0.00056577
 0.00057568 0.00058559 0.0005955  0.00060541 0.00061532 0.00062523
 0.00063514 0.00064505 0.00065495 0.00066486 0.00067477 0.00068468
 0.00069459 0.0007045  0.00071441 0.00072432 0.00073423 0.00074414
 0.00075405 0.00076396 0.00077387 0.00078378 0.00079369 0.0008036
 0.00081351 0.00082342 0.00083333 0.00084324 0.00085315 0.00086306
 0.00087297 0.00088288 0.00089279 0.0009027  0.00091261 0.00092252
 0.00093243 0.00094234 0.00095225 0.00096216 0.00097207 0.00098198
 0.00099189 0.0010018  0.00101171 0.00102162 0.00103153 0.00104144
 0.00105135 0.00106126 0.00107117 0.00108108 0.00109099 0.0011009
 0.00111081 0.00112072 0.00113063 0.00114054 0.00115045 0.00116036
 0.00117027 0.00118018 0.00119009 0.0012     0.00120991 0.00121982
 0.00122973 0.00123964 0.00124955 0.00125946 0.00126937 0.00127928
 0.00128919 0.0012991  0.00130901 0.00131892 0.00132883 0.00133874
 0.00134865 0.00135856 0.00136847 0.00137838 0.00138829 0.0013982
 0.00140811 0.00141802 0.00142793 0.00143784 0.00144775 0.00145766
 0.00146757 0.00147748 0.00148739 0.0014973  0.00150721 0.00151712
 0.00152703 0.00153694 0.00154685 0.00155676 0.00156667 0.00157658
 0.00158649 0.0015964  0.00160631 0.00161622 0.00162613 0.00163604
 0.00164595 0.00165586 0.00166577 0.00167568 0.00168559 0.0016955
 0.00170541 0.00171532 0.00172523 0.00173514 0.00174505 0.00175495
 0.00176486 0.00177477 0.00178468 0.00179459 0.0018045  0.00181441
 0.00182432 0.00183423 0.00184414 0.00185405 0.00186396 0.00187387
 0.00188378 0.00189369 0.0019036  0.00191351 0.00192342 0.00193333
 0.00194324 0.00195315 0.00196306 0.00197297 0.00198288 0.00199279
 0.0020027  0.00201261 0.00202252 0.00203243 0.00204234 0.00205225
 0.00206216 0.00207207 0.00208198 0.00209189 0.0021018  0.00211171
 0.00212162 0.00213153 0.00214144 0.00215135 0.00216126 0.00217117
 0.00218108 0.00219099 0.0022009  0.00221081 0.00222072 0.00223063
 0.00224054 0.00225045 0.00226036 0.00227027 0.00228018 0.00229009
 0.0023     0.00230991 0.00231982 0.00232973 0.00233964 0.00234955
 0.00235946 0.00236937 0.00237928 0.00238919 0.0023991  0.00240901
 0.00241892 0.00242883 0.00243874 0.00244865 0.00245856 0.00246847
 0.00247838 0.00248829 0.0024982  0.00250811 0.00251802 0.00252793
 0.00253784 0.00254775 0.00255766 0.00256757 0.00257748 0.00258739
 0.0025973  0.00260721 0.00261712 0.00262703 0.00263694 0.00264685
 0.00265676 0.00266667 0.00267658 0.00268649 0.0026964  0.00270631
 0.00271622 0.00272613 0.00273604 0.00274595 0.00275586 0.00276577
 0.00277568 0.00278559 0.0027955  0.00280541 0.00281532 0.00282523
 0.00283514 0.00284505 0.00285495 0.00286486 0.00287477 0.00288468
 0.00289459 0.0029045  0.00291441 0.00292432 0.00293423 0.00294414
 0.00295405 0.00296396 0.00297387 0.00298378 0.00299369 0.0030036
 0.00301351 0.00302342 0.00303333 0.00304324 0.00305315 0.00306306
 0.00307297 0.00308288 0.00309279 0.0031027  0.00311261 0.00312252
 0.00313243 0.00314234 0.00315225 0.00316216 0.00317207 0.00318198
 0.00319189 0.0032018  0.00321171 0.00322162 0.00323153 0.00324144
 0.00325135 0.00326126 0.00327117 0.00328108 0.00329099 0.0033009
 0.00331081 0.00332072 0.00333063 0.00334054 0.00335045 0.00336036
 0.00337027 0.00338018 0.00339009 0.0034     0.00340991 0.00341982
 0.00342973 0.00343964 0.00344955 0.00345946 0.00346937 0.00347928
 0.00348919 0.0034991  0.00350901 0.00351892 0.00352883 0.00353874
 0.00354865 0.00355856 0.00356847 0.00357838 0.00358829 0.0035982
 0.00360811 0.00361802 0.00362793 0.00363784 0.00364775 0.00365766
 0.00366757 0.00367748 0.00368739 0.0036973  0.00370721 0.00371712
 0.00372703 0.00373694 0.00374685 0.00375676 0.00376667 0.00377658
 0.00378649 0.0037964  0.00380631 0.00381622 0.00382613 0.00383604
 0.00384595 0.00385586 0.00386577 0.00387568 0.00388559 0.0038955
 0.00390541 0.00391532 0.00392523 0.00393514 0.00394505 0.00395495
 0.00396486 0.00397477 0.00398468 0.00399459 0.0040045  0.00401441
 0.00402432 0.00403423 0.00404414 0.00405405 0.00406396 0.00407387
 0.00408378 0.00409369 0.0041036  0.00411351 0.00412342 0.00413333
 0.00414324 0.00415315 0.00416306 0.00417297 0.00418288 0.00419279
 0.0042027  0.00421261 0.00422252 0.00423243 0.00424234 0.00425225
 0.00426216 0.00427207 0.00428198 0.00429189 0.0043018  0.00431171
 0.00432162 0.00433153 0.00434144 0.00435135 0.00436126 0.00437117
 0.00438108 0.00439099 0.0044009  0.00441081 0.00442072 0.00443063
 0.00444054 0.00445045 0.00446036 0.00447027 0.00448018 0.00449009
 0.0045     0.00450991 0.00451982 0.00452973 0.00453964 0.00454955
 0.00455946 0.00456937 0.00457928 0.00458919 0.0045991  0.00460901
 0.00461892 0.00462883 0.00463874 0.00464865 0.00465856 0.00466847
 0.00467838 0.00468829 0.0046982  0.00470811 0.00471802 0.00472793
 0.00473784 0.00474775 0.00475766 0.00476757 0.00477748 0.00478739
 0.0047973  0.00480721 0.00481712 0.00482703 0.00483694 0.00484685
 0.00485676 0.00486667 0.00487658 0.00488649 0.0048964  0.00490631
 0.00491622 0.00492613 0.00493604 0.00494595 0.00495586 0.00496577
 0.00497568 0.00498559 0.0049955  0.00500541 0.00501532 0.00502523
 0.00503514 0.00504505 0.00505495 0.00506486 0.00507477 0.00508468
 0.00509459 0.0051045  0.00511441 0.00512432 0.00513423 0.00514414
 0.00515405 0.00516396 0.00517387 0.00518378 0.00519369 0.0052036
 0.00521351 0.00522342 0.00523333 0.00524324 0.00525315 0.00526306
 0.00527297 0.00528288 0.00529279 0.0053027  0.00531261 0.00532252
 0.00533243 0.00534234 0.00535225 0.00536216 0.00537207 0.00538198
 0.00539189 0.0054018  0.00541171 0.00542162 0.00543153 0.00544144
 0.00545135 0.00546126 0.00547117 0.00548108 0.00549099 0.0055009
 0.00551081 0.00552072 0.00553063 0.00554054 0.00555045 0.00556036
 0.00557027 0.00558018 0.00559009 0.0056     0.00560991 0.00561982
 0.00562973 0.00563964 0.00564955 0.00565946 0.00566937 0.00567928
 0.00568919 0.0056991  0.00570901 0.00571892 0.00572883 0.00573874
 0.00574865 0.00575856 0.00576847 0.00577838 0.00578829 0.0057982
 0.00580811 0.00581802 0.00582793 0.00583784 0.00584775 0.00585766
 0.00586757 0.00587748 0.00588739 0.0058973  0.00590721 0.00591712
 0.00592703 0.00593694 0.00594685 0.00595676 0.00596667 0.00597658
 0.00598649 0.0059964  0.00600631 0.00601622 0.00602613 0.00603604
 0.00604595 0.00605586 0.00606577 0.00607568 0.00608559 0.0060955
 0.00610541 0.00611532 0.00612523 0.00613514 0.00614505 0.00615495
 0.00616486 0.00617477 0.00618468 0.00619459 0.0062045  0.00621441
 0.00622432 0.00623423 0.00624414 0.00625405 0.00626396 0.00627387
 0.00628378 0.00629369 0.0063036  0.00631351 0.00632342 0.00633333
 0.00634324 0.00635315 0.00636306 0.00637297 0.00638288 0.00639279
 0.0064027  0.00641261 0.00642252 0.00643243 0.00644234 0.00645225
 0.00646216 0.00647207 0.00648198 0.00649189 0.0065018  0.00651171
 0.00652162 0.00653153 0.00654144 0.00655135 0.00656126 0.00657117
 0.00658108 0.00659099 0.0066009  0.00661081 0.00662072 0.00663063
 0.00664054 0.00665045 0.00666036 0.00667027 0.00668018 0.00669009
 0.0067     0.00670991 0.00671982 0.00672973 0.00673964 0.00674955
 0.00675946 0.00676937 0.00677928 0.00678919 0.0067991  0.00680901
 0.00681892 0.00682883 0.00683874 0.00684865 0.00685856 0.00686847
 0.00687838 0.00688829 0.0068982  0.00690811 0.00691802 0.00692793
 0.00693784 0.00694775 0.00695766 0.00696757 0.00697748 0.00698739
 0.0069973  0.00700721 0.00701712 0.00702703 0.00703694 0.00704685
 0.00705676 0.00706667 0.00707658 0.00708649 0.0070964  0.00710631
 0.00711622 0.00712613 0.00713604 0.00714595 0.00715586 0.00716577
 0.00717568 0.00718559 0.0071955  0.00720541 0.00721532 0.00722523
 0.00723514 0.00724505 0.00725495 0.00726486 0.00727477 0.00728468
 0.00729459 0.0073045  0.00731441 0.00732432 0.00733423 0.00734414
 0.00735405 0.00736396 0.00737387 0.00738378 0.00739369 0.0074036
 0.00741351 0.00742342 0.00743333 0.00744324 0.00745315 0.00746306
 0.00747297 0.00748288 0.00749279 0.0075027  0.00751261 0.00752252
 0.00753243 0.00754234 0.00755225 0.00756216 0.00757207 0.00758198
 0.00759189 0.0076018  0.00761171 0.00762162 0.00763153 0.00764144
 0.00765135 0.00766126 0.00767117 0.00768108 0.00769099 0.0077009
 0.00771081 0.00772072 0.00773063 0.00774054 0.00775045 0.00776036
 0.00777027 0.00778018 0.00779009 0.0078     0.00780991 0.00781982
 0.00782973 0.00783964 0.00784955 0.00785946 0.00786937 0.00787928
 0.00788919 0.0078991  0.00790901 0.00791892 0.00792883 0.00793874
 0.00794865 0.00795856 0.00796847 0.00797838 0.00798829 0.0079982
 0.00800811 0.00801802 0.00802793 0.00803784 0.00804775 0.00805766
 0.00806757 0.00807748 0.00808739 0.0080973  0.00810721 0.00811712
 0.00812703 0.00813694 0.00814685 0.00815676 0.00816667 0.00817658
 0.00818649 0.0081964  0.00820631 0.00821622 0.00822613 0.00823604
 0.00824595 0.00825586 0.00826577 0.00827568 0.00828559 0.0082955
 0.00830541 0.00831532 0.00832523 0.00833514 0.00834505 0.00835495
 0.00836486 0.00837477 0.00838468 0.00839459 0.0084045  0.00841441
 0.00842432 0.00843423 0.00844414 0.00845405 0.00846396 0.00847387
 0.00848378 0.00849369 0.0085036  0.00851351 0.00852342 0.00853333
 0.00854324 0.00855315 0.00856306 0.00857297 0.00858288 0.00859279
 0.0086027  0.00861261 0.00862252 0.00863243 0.00864234 0.00865225
 0.00866216 0.00867207 0.00868198 0.00869189 0.0087018  0.00871171
 0.00872162 0.00873153 0.00874144 0.00875135 0.00876126 0.00877117
 0.00878108 0.00879099 0.0088009  0.00881081 0.00882072 0.00883063
 0.00884054 0.00885045 0.00886036 0.00887027 0.00888018 0.00889009
 0.0089     0.00890991 0.00891982 0.00892973 0.00893964 0.00894955
 0.00895946 0.00896937 0.00897928 0.00898919 0.0089991  0.00900901
 0.00901892 0.00902883 0.00903874 0.00904865 0.00905856 0.00906847
 0.00907838 0.00908829 0.0090982  0.00910811 0.00911802 0.00912793
 0.00913784 0.00914775 0.00915766 0.00916757 0.00917748 0.00918739
 0.0091973  0.00920721 0.00921712 0.00922703 0.00923694 0.00924685
 0.00925676 0.00926667 0.00927658 0.00928649 0.0092964  0.00930631
 0.00931622 0.00932613 0.00933604 0.00934595 0.00935586 0.00936577
 0.00937568 0.00938559 0.0093955  0.00940541 0.00941532 0.00942523
 0.00943514 0.00944505 0.00945495 0.00946486 0.00947477 0.00948468
 0.00949459 0.0095045  0.00951441 0.00952432 0.00953423 0.00954414
 0.00955405 0.00956396 0.00957387 0.00958378 0.00959369 0.0096036
 0.00961351 0.00962342 0.00963333 0.00964324 0.00965315 0.00966306
 0.00967297 0.00968288 0.00969279 0.0097027  0.00971261 0.00972252
 0.00973243 0.00974234 0.00975225 0.00976216 0.00977207 0.00978198
 0.00979189 0.0098018  0.00981171 0.00982162 0.00983153 0.00984144
 0.00985135 0.00986126 0.00987117 0.00988108 0.00989099 0.0099009
 0.00991081 0.00992072 0.00993063 0.00994054 0.00995045 0.00996036
 0.00997027 0.00998018 0.00999009 0.01      ]

Prueba Final con 100 propiedades :

X_100 = CABA_100[feature_cols]
y_100 = CABA_100.price_per_m2

precio_m2_predecido=lassocv.predict(X_100)
precio_m2_real=CABA_100.price_per_m2

precio_real_vs_prediccion=pd.DataFrame ({'precio_m2_real':precio_m2_real,
                                       'precio_m2_prediccion':precio_m2_predecido,
                                        'lat':CABA_100.lat,
                                        'lon':CABA_100.lon})                                    
precio_real_vs_prediccion

	precio_m2_real	precio_m2_prediccion	lat	lon
16763	2941.176471	3365.752207	-34.580504	-58.405874
11681	2260.344828	3146.229585	-34.598737	-58.426602
4417	2093.017442	1961.334965	-34.625386	-58.461752
3279	4279.069767	3484.253832	-34.594305	-58.421268
27010	2611.940299	2778.372715	-34.596020	-58.388387
...	...	...	...	...
12312	2307.692308	2806.461122	-34.556037	-58.446535
1660	3888.888889	3789.600970	-34.592188	-58.379074
5719	1932.000000	2178.070797	-34.621769	-58.422108
14224	1795.454545	2049.965511	-34.606950	-58.409365
2214	2244.897959	2098.702279	-34.631323	-58.466145

100 rows × 4 columns

precio_real_vs_prediccion['Diferencia'] = (precio_real_vs_prediccion['precio_m2_real'] - precio_real_vs_prediccion['precio_m2_prediccion'])/precio_real_vs_prediccion['precio_m2_real']

precio_real_vs_prediccion ## CUANDO LA DIFERENCIA DA NEGATIVA EL PREDECIDO ES MAS CARO QUE EL REAL

	precio_m2_real	precio_m2_prediccion	lat	lon	Diferencia
16763	2941.176471	3365.752207	-34.580504	-58.405874	-0.144356
11681	2260.344828	3146.229585	-34.598737	-58.426602	-0.391925
4417	2093.017442	1961.334965	-34.625386	-58.461752	0.062915
3279	4279.069767	3484.253832	-34.594305	-58.421268	0.185745
27010	2611.940299	2778.372715	-34.596020	-58.388387	-0.063720
...	...	...	...	...	...
12312	2307.692308	2806.461122	-34.556037	-58.446535	-0.216133
1660	3888.888889	3789.600970	-34.592188	-58.379074	0.025531
5719	1932.000000	2178.070797	-34.621769	-58.422108	-0.127366
14224	1795.454545	2049.965511	-34.606950	-58.409365	-0.141753
2214	2244.897959	2098.702279	-34.631323	-58.466145	0.065124

100 rows × 5 columns

precio_real_vs_prediccion.Diferencia.min() ## LA OPORTUNIDAD DE COMPRAR BARATO Y VENDER CARO

-0.3919246067186869

PROPS=np.linspace(0, 99,100)
PROPS

array([ 0.,  1.,  2.,  3.,  4.,  5.,  6.,  7.,  8.,  9., 10., 11., 12.,
       13., 14., 15., 16., 17., 18., 19., 20., 21., 22., 23., 24., 25.,
       26., 27., 28., 29., 30., 31., 32., 33., 34., 35., 36., 37., 38.,
       39., 40., 41., 42., 43., 44., 45., 46., 47., 48., 49., 50., 51.,
       52., 53., 54., 55., 56., 57., 58., 59., 60., 61., 62., 63., 64.,
       65., 66., 67., 68., 69., 70., 71., 72., 73., 74., 75., 76., 77.,
       78., 79., 80., 81., 82., 83., 84., 85., 86., 87., 88., 89., 90.,
       91., 92., 93., 94., 95., 96., 97., 98., 99.])

#creo la figura
fig = plt.figure(figsize=(12, 8))
# creo los axes
ax = plt.axes()
ax.set(xlabel = "Número de observaciones", ylabel = "Usd / m2", title = "Precio real Vs. Precio predicho")
#ax.legend(loc='lower right', title = "series");
# dibujo en axes:
ax.plot(PROPS, precio_real_vs_prediccion.precio_m2_real, label='precio real');
# superpongo otro gráfico en axes:
ax.plot(PROPS, precio_real_vs_prediccion.precio_m2_prediccion, label='precio predicción');
etiquetas = ['Precio real', 'Precio predicción'] 
ax.legend(etiquetas, loc='upper right', frameon=True);

#precio_real_vs_prediccion.to_csv('../Desafio2Grupal/real_vs_prediccion.csv')

Prop_para_vender = precio_real_vs_prediccion[precio_real_vs_prediccion.Diferencia < 0]

Prop_para_vender.shape 

(46, 5)

fig = px.scatter_mapbox(Prop_para_vender, lat="lat", lon="lon", zoom=9, height=600, color='Diferencia',color_continuous_scale=["red", "green", "blue"])
fig.update_layout(mapbox_style="carto-positron")
fig.update_layout(margin={"r":0,"t":0,"l":0,"b":0})
fig.show()