# Import libraries require

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

%matplotlib inline
sns.set(color_codes = True)
# Load the data into data frame according to the file where you save it

happiness = pd.read_csv("c://2019.csv")

# Let data frame to display all the data

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

# show the data
happiness

# Rename the columns to make it more easy to understand

happiness = happiness.rename(columns = {'Overall rank': 'Rank', 'Country or region': 'Country', 
                                        'GDP per capita': 'Economy (GDP per Capita)', 'Social support': 'Social Support', 
                                        'Healthy life expectancy': 'Health (Life Expectancy)', 
                                        'Freedom to make life choices': 'Freedom',
                                        'Perceptions of corruption': 'Perception of Corruption'})

#cleaning null value and inconsistent data
happiness.drop(index=86,axis=0 , inplace=True)
happiness.drop(index=96,axis=0 , inplace=True)
happiness.drop(index=98,axis=0 , inplace=True)
happiness.drop(index=111,axis=0 , inplace=True)
happiness.drop(index=117,axis=0 , inplace=True)
happiness.drop(index=119,axis=0 , inplace=True)
happiness.drop(index=125,axis=0 , inplace=True)
happiness.drop(index=134,axis=0 , inplace=True)
happiness.drop(index=147,axis=0 , inplace=True)

# Dropping the unimportant columns

happiness.drop(['Rank'], axis = 1)

# Checking if the rows containing any duplicate data or not

duplicate_rows_happiness = happiness[happiness.duplicated()]
print("number of dyplicate rows: ", duplicate_rows_happiness.shape)

# Find the null values.

print(happiness.isnull().sum())

# Drop the missing values.
happiness = happiness.dropna() 
happiness.count()

# After dropping the values

print(happiness.isnull().sum())

# Decision Tree Regression

x = pd.DataFrame(np.c_[happiness['Economy (GDP per Capita)'], happiness['Social Support'], happiness['Health (Life Expectancy)']], 
                 columns = ['Economy (GDP per Capita)','Social Support','Health (Life Expectancy)'])
y = happiness['Score']

# split the data using train_test_split
from sklearn.model_selection import train_test_split
x_train, x_test, y_train, y_test = train_test_split(x, y, test_size = 0.2, random_state=42)

# build the model
from sklearn.tree import DecisionTreeRegressor
regressor = DecisionTreeRegressor()

# Fitting Decision Tree Regression into dataset
regressor.fit(x_train, y_train)

y_pred = regressor.predict(x_test)

# test the accuracy/performance measurement using R2
print('R2: ', regressor.score(x_test, y_test))

df = pd.DataFrame({'Real Values':y_test, 'Predicted Values':y_pred})
df

# Visualising the Decision Tree Regression Results 

X_grid = np.arange(0, 10)
X_grid = X_grid.reshape((len(X_grid), 1))
plt.scatter(y_pred, y_test)
plt.title('Decision Tree Regression')
plt.xlabel('Predicted Score')
plt.ylabel('Score')
plt.show()

from sklearn import metrics
mse = metrics.mean_squared_error(y_test, y_pred)
mse

# k-fold cross validation of decision tree regression

from sklearn.model_selection import cross_val_score
regressor = DecisionTreeRegressor()

first_cv_scores = cross_val_score(regressor, x, y, cv=5, scoring='neg_mean_squared_error')
second_cv_scores = cross_val_score(regressor, x, y, cv=10, scoring='neg_mean_squared_error')
print('mean in first_cv_scores is {0:.2f} and in second_cv_scores is {1:.2f}'.format(np.mean
(first_cv_scores),
 np.mean
(second_cv_scores)))

# neg_mean_squared_error is better when close to 0.0