Healthcare Data

-----------------------------------------------------------------------------------
--------------------------------------------------
# Importing the libraries

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import seaborn as sns

# Ignore harmless warnings

import warnings
warnings.filterwarnings("ignore")

# Set to display all the columns in dataset

pd.set_option("display.max_columns", None)

# Import psql to run queries

import pandasql as psql

-----------------------------------------------------------------------------------
--------------------------------------------------
# load the Health Insurance dataset

HealthIns = pd.read_csv(r"E:\R3SPAnalytics\00 IIT KGP Hyd\00-LATA\Datasets\

Health_Ins_Expenses.csv", header=0)

# Copy to back-up file

HealthIns_BK = HealthIns.copy()

# Display the first 5 records

HealthIns.head()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the dataset information

HealthIns.info()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Displaying Duplicate values with in dataset

HealthIns_dup = HealthIns[HealthIns.duplicated(keep='last')]

# Display the duplicate records

HealthIns_dup
-----------------------------------------------------------------------------------
--------------------------------------------------
# Remove the identified duplicate records

HealthIns = HealthIns.drop_duplicates()

# Display the shape of the dataset

HealthIns.shape
-----------------------------------------------------------------------------------
--------------------------------------------------
# Re-setting the row index

HealthIns = HealthIns.reset_index(drop=True)

# Copy file to back-up file after deletion of duplicate records

HealthIns_BK2 = HealthIns.copy()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the dataset information after delection of duplicates

HealthIns.info()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the unique values of the all the variables

HealthIns.nunique()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the missing values information of variables

HealthIns.isnull().sum()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display 'Gender' categorical variable

HealthIns['Gender'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Replace 'Gender' variable and convert to integer value.

HealthIns['Gender'] = HealthIns['Gender'].str.replace('female', '0')

HealthIns['Gender'] = HealthIns['Gender'].str.replace('male', '1')
HealthIns['Gender'] = HealthIns['Gender'].astype(int)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display 'Smoker' categorical variable

HealthIns['Smoker'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Replace 'Smoker' variable and convert as a integer value.

HealthIns['Smoker'] = HealthIns['Smoker'].str.replace('no', '0')

HealthIns['Smoker'] = HealthIns['Smoker'].str.replace('yes', '1')
HealthIns['Smoker'] = HealthIns['Smoker'].astype(int)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display 'Region' categorical variable

HealthIns['Region'].value_counts()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Replace 'Region' variable and convert to integer value.

HealthIns['Region'] = HealthIns['Region'].str.replace('northeast', '0')

HealthIns['Region'] = HealthIns['Region'].str.replace('northwest', '1')
HealthIns['Region'] = HealthIns['Region'].str.replace('southeast', '2')
HealthIns['Region'] = HealthIns['Region'].str.replace('southwest', '3')
HealthIns['Region'] = HealthIns['Region'].astype(int)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the dataset information after transformation of data

HealthIns.info()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Delete variables which are not influencing the target variable

del HealthIns['Record_ID']

# Display first 5 records

HealthIns.head()
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display all the column variables

HealthIns.columns
-----------------------------------------------------------------------------------
--------------------------------------------------
# Identify the variables for scaling (MinMaxScalar)

cols = ['Age', 'BMI', 'Children','Region']

-----------------------------------------------------------------------------------
--------------------------------------------------
# Identify the independent and Target (dependent) variables

IndepVar = []
for col in HealthIns.columns:
if col != 'Expenses':
IndepVar.append(col)

TargetVar = 'Expenses'

x = HealthIns[IndepVar]
y = HealthIns[TargetVar]
-----------------------------------------------------------------------------------
--------------------------------------------------
# Split the data into train and test (random sampling)

from sklearn.model_selection import train_test_split

x_train, x_test, y_train, y_test = train_test_split(x, y, test_size=0.3,

random_state=42)

# Display the shape for train & test data

x_train.shape, x_test.shape, y_train.shape, y_test.shape

-----------------------------------------------------------------------------------
--------------------------------------------------
# Scaling the features by using MinMaxScaler

from sklearn.preprocessing import MinMaxScaler

mmscaler = MinMaxScaler(feature_range=(0, 1))

x_train[cols] = mmscaler.fit_transform(x_train[cols])
#x_train = mmscaler.fit_transform(x_train)
x_train = pd.DataFrame(x_train)

x_test[cols] = mmscaler.fit_transform(x_test[cols])
#x_test = mmscaler.fit_transform(x_test)
x_test = pd.DataFrame(x_test)
-----------------------------------------------------------------------------------
--------------------------------------------------
Multiple Regression Algorithm
-----------------------------------------------------------------------------------
--------------------------------------------------
# Train the algorithm and build the model with train dataset

from sklearn.linear_model import LinearRegression

# Create an object for regression model

ModelRGR = LinearRegression()

# Train the model with training dataset

ModelRGR.fit(x_train, y_train)

# Predict the model with test dataset

y_pred = ModelRGR.predict(x_test)

# Evaluation metrics for Regression model

from sklearn import metrics

print('Mean Absolute Error (MAE):', round(metrics.mean_absolute_error(y_test,

y_pred),3))
print('Mean Squared Error (MSE):', round(metrics.mean_squared_error(y_test,
y_pred),3))
print('Root Mean Squared Error (RMSE):',
round(np.sqrt(metrics.mean_squared_error(y_test, y_pred)),3))
print('R2_score:', round(metrics.r2_score(y_test, y_pred),6))
print('Mean Absolute Percentage Error (MAPE):',
round(metrics.mean_absolute_percentage_error(y_test, y_pred)*100,3), '%')
print('Root Mean Squared Log Error (RMSLE):',
round(np.log(np.sqrt(metrics.mean_squared_error(y_test, y_pred))),3))

# Define the function to calculate the MAPE - Mean Absolute Percentage Error

def MAPE(y_test, y_pred):

y_test, y_pred = np.array(y_test), np.array(y_pred)
return np.mean(np.abs((y_test - y_pred) / y_test)) * 100

# Evaluation of MAPE

result = MAPE(y_test, y_pred)

print('Mean Absolute Percentage Error (MAPE):', round(result, 3), '%')

# Calculate Adjusted R squared values

r_squared = round(metrics.r2_score(y_test, y_pred),6)

adjusted_r_squared = round(1 - (1-r_squared)*(len(y)-1)/(len(y)-x.shape[1]-1),6)
print('Adj R Square: ', adjusted_r_squared)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Display the Final results

Results = pd.DataFrame({'Expenses_A':y_test, 'Expenses_P':y_pred})

# Merge two Dataframes on index of both the dataframes

ResultsFinal = HealthIns_BK2.merge(Results, left_index=True, right_index=True)

# Display 10 records randomly

ResultsFinal.sample(5)
-----------------------------------------------------------------------------------
--------------------------------------------------
# Calculate the %of Error

ResultsFinal['%Error'] = round(((ResultsFinal['Expenses_A']-
ResultsFinal['Expenses_P'])/ResultsFinal['Expenses_A'])*100,3)

# Display the random 5 records

ResultsFinal.sample(5)
-----------------------------------------------------------------------------------
--------------------------------------------------