Laptop Price Prediction

import numpy as np
import pandas as pd
import matplotlib.pyplot as plt
import seaborn as sns
from sklearn.model_selection import train_test_split
import pickle
import warnings
warnings.filterwarnings("ignore")
from sklearn.compose import ColumnTransformer
from sklearn.pipeline import Pipeline
from sklearn.preprocessing import OneHotEncoder
from sklearn.metrics import mean_squared_error, r2_score, mean_absolute_error
from sklearn.linear_model import LinearRegression,Ridge,Lasso
from sklearn.neighbors import KNeighborsRegressor
from sklearn.tree import DecisionTreeRegressor
from sklearn.ensemble import RandomForestRegressor,GradientBoostingRegressor,AdaBoostRegressor,ExtraTreesRegressor
from sklearn.svm import SVR

df=pd.read_csv('laptop_data.csv')
df.head()

df.shape

df.info()

df.isnull().sum()

df.duplicated().sum()

df.drop('Unnamed: 0',axis=1,inplace=True)

df.head()

df['Ram'] = df['Ram'].str.replace('GB','')
df['Weight'] = df['Weight'].str.replace('kg','')

df.head()

print(df['Ram'].dtype)
print(df['Weight'].dtype)

df['Ram']=df['Ram'].astype('int32')
df['Weight']=df['Weight'].astype('float32')

sns.barplot(df['Ram'],df['Price'])
plt.show()
# By change in Ram size, the price varies

df.info()

df['Company'].value_counts().plot(kind='bar')
plt.show()
# Mostly we have laptop of dell, lenovo and hp

sns.barplot(x=df['Company'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()
# Razer have most expensive laptops
# laptop price alse depends upon what type of company we choose

df['TypeName'].value_counts().plot(kind='bar')
plt.show()
# Mostly we have notebook laptops

sns.barplot(x=df['TypeName'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()
# Workstation type laptops are most expensive
# laptop price also depends upon on type of laptop

sns.distplot(df['Inches'])
plt.show()
# Mostly laptops are of 16 inches

sns.scatterplot(df['Inches'],df['Price'])
plt.show()
# With the increase in icnhes, laptop prices also increases

df['ScreenResolution'].value_counts()

df['Ips']=df['ScreenResolution'].apply(lambda x:1 if 'IPS Panel' in x else 0)

df['Ips'].value_counts().plot(kind='bar')
plt.show()
# Mostly laptops have no IPS Panel

sns.barplot(df['Ips'],df['Price'])
plt.show()
# IPS Panel laptops have high price

df['Touchscreen']=df['ScreenResolution'].apply(lambda x:1 if 'Touchscreen' in x else 0)

df['Touchscreen'].value_counts().plot(kind='bar')
plt.show()
# Mostly laptops have no touchscreen

sns.barplot(df['Touchscreen'],df['Price'])
plt.show()
# Touch Screen laptops have high price

res=df['ScreenResolution'].str.split('x')

df['X_res']=0
df['Y_res']=0
for i in range(len(res)):
    df['Y_res'][i]=res[i][1]
    df['X_res'][i]=res[i][0][-4:]

df.head()

df['X_res']=df['X_res'].astype('int32')
df['Y_res']=df['Y_res'].astype('int32')

sns.distplot(df['Y_res'])
plt.show()

sns.barplot(x=df['Y_res'],y=df['Price'])
plt.show()

sns.distplot(df['X_res'])
plt.show()

sns.barplot(x=df['X_res'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()

df.corr()['Price']

df['ppi'] = (((df['X_res']**2) + (df['Y_res']**2))**0.5)/df['Inches'].astype('float')
df.corr()['Price']
# Having strong corelation of ppi with price

df.drop(['ScreenResolution','X_res','Y_res','Inches'],axis=1,inplace=True)

df.head()

df['Cpu Brand']=df['Cpu']
for i,j in zip(range(len(df['Cpu'])),df['Cpu']):
    if 'Intel Core i3' in j:
        df['Cpu Brand'][i]='Intel Core i3'
    elif 'Intel Core i5' in j:
        df['Cpu Brand'][i]='Intel Core i5'
    elif 'Intel Core i7' in j:
        df['Cpu Brand'][i]='Intel Core i7'
    elif 'AMD' in j:
        df['Cpu Brand'][i]='AMD'
    else:
        df['Cpu Brand'][i]='Other Processor'

df['Cpu Brand'].value_counts().plot(kind='bar')
plt.show()

sns.barplot(x=df['Cpu Brand'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()

df.drop('Cpu',axis=1,inplace=True)

sns.distplot(df['Ram'])
plt.show()

sns.barplot(x=df['Ram'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()

df['Memory'].value_counts()

df['SSD']=0
df['HDD']=0
df['Hybrid']=0
df['Flash Storage']=0
df['Memory']=df['Memory'].str.replace('1TB','1000GB')
df['Memory']=df['Memory'].str.replace('1.0TB','1000GB')
df['Memory']=df['Memory'].str.replace('2.0TB','2000GB')
df['Memory']=df['Memory'].str.replace('2TB','2000GB')
for i in range(len(df)):
    a=df['Memory'][i].split('+')
    b=a[0].split('GB')
    if 'SSD' in a[0]:
        df['SSD'][i]=b[0]
    elif 'HDD' in a[0]:
        df['HDD'][i]=b[0]
    elif 'Hybrid' in a[0]:
        df['Hybrid'][i]=b[0]
    elif 'Flash Storage' in a[0]:
        df['Flash Storage'][i]=b[0]
    if '+'in df['Memory'][i]:
        c=a[1].split('GB')
        if 'SSD' in a[1]:
            df['SSD'][i]=int(df['SSD'][i])+int(c[0])
        elif 'HDD' in a[1]:
            df['HDD'][i]=int(df['HDD'][i])+int(c[0])
        elif 'Hybrid' in a[1]:
            df['Hybrid'][i]=int(df['Hybrid'][i])+int(c[0])
        elif 'Flash Storage' in a[1]:
            df['Flash Storage'][i]=int(df['Flash Storage'][i])+int(c[0])

sns.distplot(df['HDD'])
plt.show()

sns.distplot(df['SSD'])
plt.show()

sns.distplot(df['Flash Storage'])
plt.show()

sns.distplot(df['Hybrid'])
plt.show()

df.drop('Memory',axis=1,inplace=True)

df.corr()['Price']

df['Hybrid'].value_counts()

df['Flash Storage'].value_counts()

df.drop(['Hybrid','Flash Storage'],axis=1,inplace=True)

df.head()

df['Gpu Brand']=0
for i in range(len(df['Gpu'])):
    df['Gpu Brand'][i]=df['Gpu'][i].split(' ')[0]

df=df[df['Gpu Brand']!='ARM']
df.reset_index(inplace=True)

df.drop('Gpu',axis=1,inplace=True)
df.head()

df['Gpu Brand'].value_counts().plot(kind='bar')
plt.show()

sns.barplot(df['Gpu Brand'],df['Price'])
plt.show()

df['OpSys'].value_counts()

sns.barplot(df['OpSys'],df['Price'])
plt.xticks(rotation='vertical')
plt.show()

df['OS']=0
for i in range(len(df['OpSys'])):
    if 'Windows' in df['OpSys'][i]:
        df['OS'][i]='Windows'
    elif 'mac' in df['OpSys'][i]:
        df['OS'][i]='Mac'
    elif 'Mac' in df['OpSys'][i]:
        df['OS'][i]='Mac'
    else:
        df['OS'][i]='Others/No Linus/No OS'

df.drop(columns=['OpSys'],inplace=True)

df.head()

sns.barplot(x=df['OS'],y=df['Price'])
plt.xticks(rotation='vertical')
plt.show()

df['OS'].value_counts()

df.corr()['Price']

plt.figure(figsize=(10,5))
sns.heatmap(df.corr(),annot=True)
plt.show()

sns.distplot(np.log(df['Price']))
plt.show()

df['Price']=np.log(df['Price'])

df.drop('index',axis=1,inplace=True)

df.head()

X = df.drop(columns=['Price'])
y = np.log(df['Price'])

X_train,X_test,y_train,y_test = train_test_split(X,y,test_size=0.15,random_state=2)

"""### Linear regression"""

step1 = ColumnTransformer(transformers=[
    ('col_tnf',OneHotEncoder(sparse=False,drop='first'),[0,1,7,10,11])
    # We will not count price that's why our counting change
],remainder='passthrough')

step2 = LinearRegression()

pipe = Pipeline([
    ('step1',step1),
    ('step2',step2)
])

pipe.fit(X_train,y_train)

y_pred = pipe.predict(X_test)

print('R2 score',r2_score(y_test,y_pred))
print('MAE',mean_absolute_error(y_test,y_pred))