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

# path of data
path = 'https://s3-api.us-geo.objectstorage.softlayer.net/cf-courses-data/CognitiveClass/DA0101EN/automobileEDA.csv'
df = pd.read_csv(path)
df.head()


# SIMPLE LINEAR REGRESSION

from sklearn.linear_model import LinearRegression

# create the linear regression object
lm = LinearRegression()

# highway-mpg for predict price
X = df[['highway-mpg']]
Y = df[['price']]
lm.fit(X,Y)

# predict for first 5 raw of the table
lm.predict(X[0:5])

# value of intercept and coefficient
lm.intercept_
lm.coef_


# MULTIPLE LINEAR REGRESSION

Z = df[['horsepower', 'curb-weight', 'engine-size', 'highway-mpg']]

lm.fit(Z, df['price'])


# MODEL EVALUATION USING VISUALIZATION

import seaborn as sns

# simple linear regression
width = 12
height = 10
plt.figure(figsize=(width, height))
sns.regplot(x="highway-mpg", y="price", data=df)
plt.ylim(0,)     # y axis starts from 0

# Is more correlated with price "peak-rpm" or "highway-mpg"?
df[["peak-rpm","highway-mpg","price"]].corr()     # highway-mpg is more correlated

# Residual Plot
# A residual plot is a graph that shows the residuals on the vertical y-axis and the independent variable on the horizontal x-axis.
width = 12
height = 10
plt.figure(figsize=(width, height))
sns.residplot(df['highway-mpg'], df['price'])
plt.show()

# We can see from this residual plot that the residuals are not randomly spread around the x-axis, which leads us to believe that maybe a non-linear model is more appropriate for this data.


# multiple linear regression
# distribution plot: fitted values vs actual values
Y_hat = lm.predict(Z)    # make the prediction

plt.figure(figsize=(width, height))
ax1 = sns.distplot(df['price'], hist=False, color="r", label="Actual Value")
sns.distplot(Y_hat, hist=False, color="b", label="Fitted Values" , ax=ax1)
plt.title('Actual vs Fitted Values for Price')
plt.xlabel('Price (in dollars)')
plt.ylabel('Proportion of Cars')
plt.show()
plt.close()

# fitted reasonable close to actual, but there is space for improvement

# POLYNOMIAL REGRESSION

# function for plotting the data
def PlotPolly(model, independent_variable, dependent_variabble, Name):
    x_new = np.linspace(15, 55, 100)
    y_new = model(x_new)
    plt.plot(independent_variable, dependent_variabble, '.', x_new, y_new, '-')
    plt.title('Polynomial Fit with Matplotlib for Price ~ Length')
    ax = plt.gca()
    ax.set_facecolor((0.898, 0.898, 0.898))
    fig = plt.gcf()
    plt.xlabel(Name)
    plt.ylabel('Price of Cars')
    plt.show()
    plt.close()

# variables
x = df['highway-mpg']
y = df['price']

# Here we use a polynomial of the 3rd order (cubic)
f = np.polyfit(x, y, 3)
p = np.poly1d(f)
print(p)
PlotPolly(p, x, y, 'highway-mpg')

# coefficients
np.polyfit(x, y, 3)    # y = a + b1X + b2X² + b3X³

# Multivariate Polynomial function: with four X variables we will have 15 coefficient
from sklearn.preprocessing import PolynomialFeatures
pr=PolynomialFeatures(degree=2)
pr
Z_pr=pr.fit_transform(Z)
Z.shape     # The original data is of 201 samples and 4 features
Z_pr.shape     # after the transformation, there 201 samples and 15 features


# MEASURES FOR IN-SAMPLE EVALUATION

# simple linear regression
lm.fit(X, Y)     #highway_mpg_fit
lm.score(X, Y)     

# R^2: 0.4965911884339176

Yhat=lm.predict(X)     # predictions
from sklearn.metrics import mean_squared_error
mean_squared_error(df['price'], Yhat)

# MSE: 31635042.944639888

# multiple linear regression
lm.fit(Z, df['price'])     # fit the model
lm.score(Z, df['price'])     

# R^2: 0.8093562806577457

Y_predict_multifit = lm.predict(Z)     # predictions
mean_squared_error(df['price'], Y_predict_multifit)     

# MSE: 11980366.87072649

# polynomial linear regression
from sklearn.metrics import r2_score     # we use a different function
r2_score(y, p(x))     

# R^2: 0.674194666390652
mean_squared_error(df['price'], p(x))     

# MSE: 20474146.426361218

# Comparing these three models, we conclude that the MLR model is the best model to be able to predict price from our dataset.