IndianHousingAnalysis By Ahmad Raza
import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
df = pd.read_csv('IndianHouses.csv')
df.head()
Area BHK Bathroom Furnishing Locality Parking Price Status Transaction Type Per_Sqft
Semi-
0 800.0 3 2.0 Rohini Sector 25 1.0 6500000 Ready_to_move New_Property Builder_Floor NaN
Furnished
Semi- J R Designers Floors,
1 750.0 2 2.0 1.0 5000000 Ready_to_move New_Property Apartment 6667.0
Furnished Rohini Sector 24
Citizen Apartment, Rohini
2 950.0 2 2.0 Furnished 1.0 15500000 Ready_to_move Resale Apartment 6667.0
Sector 13
Semi-
3 600.0 2 2.0 Rohini Sector 24 1.0 4200000 Ready_to_move Resale Builder_Floor 6667.0
Furnished
Semi- Rohini Sector 24 carpet
4 650.0 2 2.0 1.0 6200000 Ready_to_move New_Property Builder_Floor 6667.0
Furnished area 650 sqft status R...
Data Preprocessing Part 1
df.select_dtypes(include='object').nunique()
Furnishing 3
Locality 365
Status 2
Transaction 2
Type 2
dtype: int64
df['Locality']
0 Rohini Sector 25
1 J R Designers Floors, Rohini Sector 24
2 Citizen Apartment, Rohini Sector 13
3 Rohini Sector 24
4 Rohini Sector 24 carpet area 650 sqft status R...
...
1254 Chittaranjan Park
1255 Chittaranjan Park
1256 Chittaranjan Park
1257 Chittaranjan Park Block A
1258 Chittaranjan Park
Name: Locality, Length: 1259, dtype: object
#we dont need deep information of locality we need main detail as i split in feature
df['Locality'].str.split(' ').str[1]
0 Sector
1 R
2 Apartment,
3 Sector
4 Sector
...
1254 Park
1255 Park
1256 Park
1257 Park
1258 Park
Name: Locality, Length: 1259, dtype: object
df['Locality'] = df['Locality'].str.split(' ').str[1]
#using split string funtion we split and got less number of unique values but we need top 30
plt.figure(figsize=(10,5))
df['Locality'].value_counts().head(50).plot(kind='bar')
plt.show()
df['Locality'].nunique()
119
#the number opf unique values is less than nefore but we cant handle it so we drop it
df.drop('Locality',axis=1,inplace=True)
Handle Missing Values
df.isnull().sum()
Area 0
BHK 0
Bathroom 2
Furnishing 5
Parking 33
Price 0
Status 0
Transaction 0
Type 5
Per_Sqft 241
dtype: int64
#we fill nan values by mean because of numerical values
df['Per_Sqft'] = df['Per_Sqft'].fillna(df['Per_Sqft'].mean())
#we fill nan values by mode because of categorical values
df['Bathroom'] = df['Bathroom'].fillna(df['Bathroom'].mode()[0])
df['Parking'] = df['Parking'].fillna(df['Parking'].mode()[0])
df['Furnishing'] = df['Furnishing'].fillna(df['Furnishing'].mode()[0])
df['Type'] = df['Type'].fillna(df['Type'].mode()[0])
Handle DataTypes
df.info()
<class 'pandas.core.frame.DataFrame'>
RangeIndex: 1259 entries, 0 to 1258
Data columns (total 10 columns):
# Column Non-Null Count Dtype
--- ------ -------------- -----
0 Area 1259 non-null float64
1 BHK 1259 non-null int64
2 Bathroom 1259 non-null float64
3 Furnishing 1259 non-null object
4 Parking 1259 non-null float64
5 Price 1259 non-null int64
6 Status 1259 non-null object
7 Transaction 1259 non-null object
8 Type 1259 non-null object
9 Per_Sqft 1259 non-null float64
dtypes: float64(4), int64(2), object(4)
memory usage: 98.5+ KB
df['Bathroom'] = df['Bathroom'].astype('int')
df['Parking'] = df['Parking'].astype('int')
Exploratary Data Analysis
cat_vars = ['Furnishing','Status','Transaction','Type']
num_cols = len(cat_vars)
fig , axs = plt.subplots(nrows=2,ncols=2,figsize=(15,10))
axs = axs.flatten()
for i , var in enumerate (cat_vars):
sns.countplot(y=var,data=df,ax=axs[i])
axs[i].set_title(var)
if num_cols < len(axs):
for i in range(num_cols, len(axs)):
fig.delaxes(axs[i])
fig.tight_layout()
plt.show()
int_vars = df.select_dtypes(include=['int','float'])
num_cols = len(int_vars)
fig , axs = plt.subplots(nrows=3,ncols=2,figsize=(15,10))
axs = axs.flatten()
for i , var in enumerate (int_vars):
df[var].plot(kind='hist',ax=axs[i])
axs[i].set_title(var)
if num_cols < len(axs):
for i in range(num_cols, len(axs)):
fig.delaxes(axs[i])
fig.tight_layout()
plt.show()
#exploratary data anaylsis with kde(kernal density estimation)
num = df.select_dtypes(include=['int','float']).columns.tolist()
col = len(num)
fig , axs = plt.subplots(nrows=col,ncols=2,figsize=(15,20))
axs = axs.flatten()
for i , var in enumerate (num):
sns.histplot(data=df,x=var,kde=True,ax=axs[i])
axs[i].set_title(var)
if col < len(axs):
for i in range(col, len(axs)):
fig.delaxes(axs[i])
fig.tight_layout()
plt.show()
#exploratary data anaylsis with boxplot to indentity outliers
num = df.select_dtypes(include=['int','float']).columns.tolist()
col = len(num)
fig , axs = plt.subplots(nrows=col,ncols=2,figsize=(15,20))
axs = axs.flatten()
for i , var in enumerate (num):
sns.boxplot(data=df,x=var,ax=axs[i])
axs[i].set_title(var)
if col < len(axs):
for i in range(col, len(axs)):
fig.delaxes(axs[i])
fig.tight_layout()
plt.show()
#eda using dependent feature price
cat = ['Furnishing','Status','Transaction','Type']
col = len(cat)
fig, axs = plt.subplots(nrows=col,ncols=2,figsize=(15,15))
axs = axs.flatten()
for i, var in enumerate (cat):
sns.barplot(x='Price', y=var, data=df, ax=axs[i])
axs[i].set_title(var)
if col < len(axs):
for i in range(col, len(axs)):
fig.delaxes(axs[i])
fig.tight_layout()
plt.show()
Data Preprocessing Part 2
#print all the unique values ob object datatypes to enoded them
#print all the unique values ob object datatypes to enoded them
for col in df.select_dtypes(include='object').columns:
print(f'{col}: {df[col].unique()}')
Furnishing: ['Semi-Furnished' 'Furnished' 'Unfurnished']
Status: ['Ready_to_move' 'Almost_ready']
Transaction: ['New_Property' 'Resale']
Type: ['Builder_Floor' 'Apartment']
#encoded all values by labelencoder
from sklearn import preprocessing
for col in df.select_dtypes(include=['object']).columns:
label_encoder = preprocessing.LabelEncoder()
label_encoder.fit(df[col].unique())
df[col] = label_encoder.transform(df[col])
print(f'{col} : {df[col].unique()}')
Furnishing : [1 0 2]
Status : [1 0]
Transaction : [0 1]
Type : [1 0]
#co_relation heatmap
plt.figure(figsize=(15,10))
sns.heatmap(df.corr(),fmt='.2g',annot=True)
plt.show()
Loading [MathJax]/jax/output/CommonHTML/fonts/TeX/fontdata.js
