X23 Group 4

Project with Real Estate Dataset

Christian Rodriguez, Emmanuel Salcedo, Ranjita Summan,
Crystian Chavez
In [10]: import sys
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
from matplotlib import rcParams
import seaborn as sns
from scipy.stats import zscore
from sklearn.linear_model import LinearRegression
from sklearn.preprocessing import PolynomialFeatures
from sklearn.metrics import mean_squared_error, r2_score
from sklearn.model_selection import train_test_split, cross_val_score

# allow output to span multiple output lines in the console

pd.set_option('display.max_columns', 500)

Dataset Selection: We have chosen the USA Real Estate Dataset from Kaggle, which contains
information about real estate properties in the USA.

https://www.kaggle.com/datasets/ahmedshahriarsakib/usa-real-estate-dataset?
resource=download

Google Drive is mounted to colab and placed into path /content/drive/MyDrive/realtor-

data.csv.

In [11]: #from google.colab import drive

#drive.mount('/content/drive')

Goal Definition: Goal is to build a system that predicts the cost of a home based on features
such as footage, number of beds, and baths, on a per-state basis.

Feature Selection:In this case, the relevant features are "footage," "beds," and "baths." These
features will be used to predict the cost of a home.

Preprocessing: Perform preprocessing steps on the dataset to prepare it for analysis. This may
include handling missing values, encoding categorical variables, and scaling numerical features.
Skipping this for now.

Exploration and Visualization: Here we are exploring the relationship between variables by using
plots such as histograms, scatter plots, or box plots.
In [12]: #df = pd.read_csv('/content/drive/MyDrive/realtor-data.csv')
#df.head(100)
df = pd.read_csv(r'C:\Users\Crystian\Documents\GitHub\CST383\realtor-data.csv')
df.head(20)
Out[12]: status bed bath acre_lot city state zip_code house_size prev_sold_date price

Puerto
0 for_sale 3.0 2.0 0.12 Adjuntas 601.0 920.0 NaN 105000.0
Rico

Puerto
1 for_sale 4.0 2.0 0.08 Adjuntas 601.0 1527.0 NaN 80000.0
Rico

Juana Puerto
2 for_sale 2.0 1.0 0.15 795.0 748.0 NaN 67000.0
Diaz Rico

Puerto
3 for_sale 4.0 2.0 0.10 Ponce 731.0 1800.0 NaN 145000.0
Rico

Puerto
4 for_sale 6.0 2.0 0.05 Mayaguez 680.0 NaN NaN 65000.0
Rico

San Puerto
5 for_sale 4.0 3.0 0.46 612.0 2520.0 NaN 179000.0
Sebastian Rico

Puerto
6 for_sale 3.0 1.0 0.20 Ciales 639.0 2040.0 NaN 50000.0
Rico

Puerto
7 for_sale 3.0 2.0 0.08 Ponce 731.0 1050.0 NaN 71600.0
Rico

Puerto
8 for_sale 2.0 1.0 0.09 Ponce 730.0 1092.0 NaN 100000.0
Rico

Las Puerto
9 for_sale 5.0 3.0 7.46 670.0 5403.0 NaN 300000.0
Marias Rico

Puerto
10 for_sale 3.0 2.0 13.39 Isabela 662.0 1106.0 NaN 89000.0
Rico

Juana Puerto
11 for_sale 3.0 2.0 0.08 795.0 1045.0 NaN 150000.0
Diaz Rico

Puerto
12 for_sale 3.0 2.0 0.10 Lares 669.0 4161.0 NaN 155000.0
Rico

Puerto
13 for_sale 5.0 2.0 0.12 Utuado 641.0 1620.0 NaN 79000.0
Rico

Puerto
14 for_sale 5.0 5.0 0.74 Ponce 731.0 2677.0 NaN 649000.0
Rico

Puerto
15 for_sale 3.0 2.0 0.08 Yauco 698.0 1100.0 NaN 120000.0
Rico

Puerto
16 for_sale 4.0 4.0 0.22 Mayaguez 680.0 3450.0 NaN 235000.0
Rico

Puerto
17 for_sale 3.0 2.0 0.08 Ponce 728.0 1500.0 NaN 105000.0
Rico

San Puerto
18 for_sale 3.0 2.0 3.88 685.0 4000.0 NaN 575000.0
Sebastian Rico

Puerto
19 for_sale 6.0 3.0 0.25 Anasco 610.0 1230.0 NaN 140000.0
Rico
 New Section
In [13]: df.info()

<class 'pandas.core.frame.DataFrame'>
RangeIndex: 407890 entries, 0 to 407889
Data columns (total 10 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 status 407890 non-null object
1 bed 320108 non-null float64
2 bath 321618 non-null float64
3 acre_lot 331873 non-null float64
4 city 407838 non-null object
5 state 407890 non-null object
6 zip_code 407693 non-null float64
7 house_size 324365 non-null float64
8 prev_sold_date 140950 non-null object
9 price 407890 non-null float64
dtypes: float64(6), object(4)
memory usage: 31.1+ MB

In [14]: df.describe()

Out[14]: bed bath acre_lot zip_code house_size price

count 320108.000000 321618.000000 331873.000000 407693.000000 3.243650e+05 4.078900e+05

mean 3.500200 2.566545 17.418487 3299.396838 2.222783e+03 6.758307e+05

std 2.320135 2.391618 931.723094 2222.641467 3.333098e+03 1.178266e+06

min 1.000000 1.000000 0.000000 601.000000 1.000000e+02 1.000000e+00

25% 2.000000 2.000000 0.200000 1890.000000 1.206000e+03 1.999000e+05

50% 3.000000 2.000000 0.560000 2822.000000 1.767000e+03 3.979000e+05

75% 4.000000 3.000000 2.200000 4630.000000 2.640000e+03 7.090000e+05

max 99.000000 198.000000 100000.000000 99999.000000 1.450112e+06 6.000000e+07

Running checks to see what columns have data missing and calculating the percentage in
relation to total.

In [15]: missing_total = df.isna().sum()

print("# of items missing from columns")
print(missing_total)
print(" ")
missing_total_per = df.isna().sum()*100/len(df)
print("Percentage of items missing from columns")
print(missing_total_per)
 # of items missing from columns
status 0
bed 87782
bath 86272
acre_lot 76017
city 52
state 0
zip_code 197
house_size 83525
prev_sold_date 266940
price 0
dtype: int64

Percentage of items missing from columns

status 0.000000
bed 21.520998
bath 21.150800
acre_lot 18.636642
city 0.012749
state 0.000000
zip_code 0.048297
house_size 20.477335
prev_sold_date 65.444115
price 0.000000
dtype: float64

We will drop rows that have items missing from bed and bath, and house size to trim data to
have homes with a price, bed and bath numbers.

In [16]: df = df.dropna(subset=['bed', 'bath', 'house_size', 'zip_code'])

# Resetting the indices using df.reset_index()

df = df.reset_index(drop=True)

In [17]: missing_total2 = df.isna().sum()

print("# of items missing from columns")
print(missing_total2)

# of items missing from columns

status 0
bed 0
bath 0
acre_lot 73083
city 1
state 0
zip_code 0
house_size 0
prev_sold_date 183947
price 0
dtype: int64

Creating a simple scatter plot to view how the number of beds relates to the cost of a home.
(Price of home in scientific notation)

In [18]: sns.scatterplot(data=df, x='bed', y='price', color='darkred')

plt.ticklabel_format(style='plain')
plt.title("Number of beds compared to price of homes")
 Text(0.5, 1.0, 'Number of beds compared to price of homes')
Out[18]:

Another scatterplot of number of baths related to price of home. (scientific notation)

In [19]: sns.scatterplot(data=df, x='bath', y='price', color='green')

plt.ticklabel_format(style='plain')
 Looking at the scatter plots we can see that the numbers are skewed quite a bit on the number
of beds and bath, we will continue to clean up the dataset and remove homes that have more
than 8 bedrooms and 10 bath so that we can get a better representation of the average home
on the market.

In [20]: indexBed = df[(df['bed'] > 10)].index

df.drop(indexBed , inplace=True)

indexBath = df[(df['bath'] > 8)].index

df.drop(indexBath , inplace=True)

indexPrice = df[(df['price'] > 999999)].index

df.drop(indexPrice , inplace=True)

df.head(15)
Out[20]: status bed bath acre_lot city state zip_code house_size prev_sold_date price

Puerto
0 for_sale 3.0 2.0 0.12 Adjuntas 601.0 920.0 NaN 105000.0
Rico

Puerto
1 for_sale 4.0 2.0 0.08 Adjuntas 601.0 1527.0 NaN 80000.0
Rico

Juana Puerto
2 for_sale 2.0 1.0 0.15 795.0 748.0 NaN 67000.0
Diaz Rico

Puerto
3 for_sale 4.0 2.0 0.10 Ponce 731.0 1800.0 NaN 145000.0
Rico

San Puerto
4 for_sale 4.0 3.0 0.46 612.0 2520.0 NaN 179000.0
Sebastian Rico

Puerto
5 for_sale 3.0 1.0 0.20 Ciales 639.0 2040.0 NaN 50000.0
Rico

Puerto
6 for_sale 3.0 2.0 0.08 Ponce 731.0 1050.0 NaN 71600.0
Rico

Puerto
7 for_sale 2.0 1.0 0.09 Ponce 730.0 1092.0 NaN 100000.0
Rico

Las Puerto
8 for_sale 5.0 3.0 7.46 670.0 5403.0 NaN 300000.0
Marias Rico

Puerto
9 for_sale 3.0 2.0 13.39 Isabela 662.0 1106.0 NaN 89000.0
Rico

Juana Puerto
10 for_sale 3.0 2.0 0.08 795.0 1045.0 NaN 150000.0
Diaz Rico

Puerto
11 for_sale 3.0 2.0 0.10 Lares 669.0 4161.0 NaN 155000.0
Rico

Puerto
12 for_sale 5.0 2.0 0.12 Utuado 641.0 1620.0 NaN 79000.0
Rico

Puerto
13 for_sale 5.0 5.0 0.74 Ponce 731.0 2677.0 NaN 649000.0
Rico

Puerto
14 for_sale 3.0 2.0 0.08 Yauco 698.0 1100.0 NaN 120000.0
Rico

In [21]: sns.scatterplot(data=df, x='bed', y='price', color='darkred')

plt.ticklabel_format(style='plain')
plt.title("Number of beds compared to price of homes")

Text(0.5, 1.0, 'Number of beds compared to price of homes')

Out[21]:
In [22]: sns.scatterplot(data=df, x='bath', y='price', color='green')
plt.ticklabel_format(style='plain')
 Using seaborn count plot to see the number of homes for sale per state. As we found, some
states do not have homes for sale but just land only.

In [23]: sns.countplot(data=df, x='state')

plt.xticks(rotation=80)
plt.title("Number of homes for sale by state")

Text(0.5, 1.0, 'Number of homes for sale by state')

Out[23]:

In [24]: X = df[['bed']].values
y = df[['price']].values

reg = LinearRegression()
reg.fit(X,y)

sns.scatterplot(data=df, x='bed', y='price')

plt.title("Linear model of Price using beds")

plt.plot(X, reg.intercept_ + reg.coef_*X, linestyle='dashed', color='black')

plt.ticklabel_format(style='plain')
In [25]: print(f'intercept: {reg.intercept_[0]:.2f}')
print(f'coefficient for price: {reg.coef_[0]}')
print(f'r-sqaured value: {reg.score(X,y):.2f}')

intercept: 401245.26
coefficient for price: [11307.83415314]
r-sqaured value: 0.00

In [26]: predictors = ['bed', 'bath']

X = df[predictors]
y = df['price']

reg2 = LinearRegression()
reg2.fit(X,y)

print(f'intercept: {reg2.intercept_:.2f}')
print('coefficients:')
for i, coef in enumerate(reg2.coef_):
print(f' {predictors[i]}: {coef:.2f}')

intercept: 290364.02
coefficients:
bed: -31057.80
bath: 112282.24

In [27]: matrix = pd.DataFrame({'bed': [3], 'bath': [3]})

results = reg2.predict(matrix)[0]
print(f'{results:.2f}')
 534037.32

In [28]: def plot_actual_predicted(actual, predicted, title):

sns.scatterplot(x=actual, y=predicted)
a = (np.min(actual),np.min(predicted))
b = (np.max(actual),np.max(predicted))

plt.plot((a,b), (a,b), linestyle='dashed', color='black')

plt.title(title)
plt.xlabel('actual')
plt.ylabel('predicted')
plt.ticklabel_format(style='plain')

In [29]: plot_actual_predicted(y, reg2.predict(X), 'Predicted by actual (Home price predictions

In [30]: predictors = ['bed', 'bath', 'zip_code']

X = df[predictors]
y = df['price']

reg3 = LinearRegression()
reg3.fit(X,y)

print(f'intercept: {reg3.intercept_:.2f}')
print('coefficients:')
for i, coef in enumerate(reg3.coef_):
print(f' {predictors[i]}: {coef:.2f}')
intercept: 347289.66
coefficients:
bed: -30941.97
bath: 114300.17
zip_code: -18.78