import matplotlib.pyplot as plt
import numpy as np 
from sklearn.linear_model import LinearRegression
import pandas as pd

houses_full=pd.read_csv('dados/Melbourne_housing_FULL.csv')
houses=houses_full.dropna()
houses=houses.loc[houses['BuildingArea']<1500]
houses=houses.sample(1000,random_state=123)
predictors=['BuildingArea','Rooms','Distance']

houses[predictors].head()

x=np.array(houses.loc[:,predictors])
y=np.array(houses.loc[:,'Price'])

def priceshrink(v,t):
    return int(v/1000000)

fig, sc = plt.subplots(1,len(predictors))
fig.suptitle('Scatter plots between predictors and target')
for i in range(len(predictors)):
    sc[i].scatter(x[:,i], y)
    sc[i].set(xlabel=predictors[i])
    #sc[i].yaxis.set_major_formatter(plt.NullFormatter())
    sc[i].yaxis.set_major_formatter(plt.FuncFormatter(priceshrink))
sc[0].set(ylabel='Price')
#sc[0].yaxis.set_major_formatter(plt.FuncFormatter(priceshrink))
plt.show()

model = LinearRegression(fit_intercept=True)


model.fit(x[:,0:1], y)

xfit = np.linspace(0, max(x[:,0]), 1000)
# newaxis gives xfit the shape of a matrix
yfit = model.predict(xfit[:,np.newaxis])

ax=plt.axes(label='lrfit')
ax.scatter(x[:,0], y)
ax.set_ylabel('Price')
ax.set_xlabel('BuildingArea')
ax.set_xlim(left=0)
ax.set_xlim(left=0,right=max(x[:,0]))
ax.yaxis.set_major_formatter(plt.FuncFormatter(priceshrink))
ax.plot(xfit, yfit,color='red')
plt.show()

print("Model slope:    ", model.coef_[0])
print("Model intercept:", model.intercept_)

print("Model R2:    ", model.score(x[:,0:1],y))

model.fit(x, y)
print("Predictors:",predictors)
print("Coefficients (alphas):    ", model.coef_)
print("Model intercept:", model.intercept_)
print("Model R2:    ", model.score(x,y))

rng = np.random.RandomState(1)
x = 10 * rng.rand(50)
y = (np.sin(x) + 0.1 * rng.randn(50) +5)*24

ax=plt.axes(label='lrknn')
ax.set_xlim(0,10)
ax.scatter(x, y)

# KNN

from sklearn.neighbors import KNeighborsRegressor
neigh = KNeighborsRegressor(n_neighbors=2)
neigh.fit(x[:,np.newaxis], y)
yfitn=neigh.predict(xfit[:,np.newaxis])

ax=plt.axes(label='lrknn')
ax.set_xlim(0,10)
ax.scatter(x, y)
ax.plot(xfit, yfitn,color='green')
plt.show()