# Importez les modules nécessaires.
import pandas as pd 
import geopandas as gp 
import matplotlib.pyplot as plt
import numpy as np
from mpl_toolkits.axes_grid1 import make_axes_locatable
from koala import Koala

# Importez les bases de données.
uvv = pd.read_excel("https://gis.cancer.gov/tools/uv-exposure/uv-county.xlsx")
all_mela = pd.read_csv("melanome.csv", skiprows=2)
white_mela = pd.read_csv("melanome_blanc.csv", skiprows=2)

# La base geojson des états d'amérique
usa = gp.read_file("usastates.geojson")

# Modification de la base UV (DataFrame `uvv`)
uv = pd.DataFrame({"state":(), "uvrate":()})
state = uvv.STATENAME.drop_duplicates()

for i in state:
    x = uvv[uvv.STATENAME == i]
    m = x.mean()
    r = pd.DataFrame([(i, m[1])], columns=("state", "uvrate"))
    uv = uv.append(r)

uv.index = range(49)

# Modification des bases mélanome
# Pour la base toutes ethnies
cc = all_mela.drop([49, 50, 51, 52, 53])
cc = cc.sort_values(by=['United States'])
cc.index = range(49)
cc = cc.loc[:, ['United States', 'Crude Rate', 'Age-adjusted Rate']]
cc.columns = ['state', 'Crude Rate', 'Age-adjusted Rate']

# Pour la base blancs non hispaniques
ccw = white_mela.drop([49, 50, 51, 52])
ccw = ccw.sort_values(by=['United States'])
ccw.index = range(49)
ccw = ccw.loc[:, ['United States', 'Crude Rate', 'Age-adjusted Rate']]
ccw.columns = ['state', 'Crude Rate W', 'Age-adjusted Rate W']

# Assemblage des 3 bases
ccuv = cc.join(uv.set_index('state'), on='state')
ccuv = ccuv.join(ccw.set_index('state'), on='state')

# Modification du GeoDataFrame usa states
x = list(usa.columns)
x[6] = 'state'
usa.columns = x
x.remove('state')
x.remove('geometry')
usa = usa.drop(columns=x)

ccuvusa = usa.join(ccuv.set_index('state'), on='state')
ccuvusa = ccuvusa.drop([7, 17, 34])
ccuvusa = ccuvusa.drop([10, 14, 46])

ccuvusa.to_file("baseUV.geojson", driver='GeoJSON')

# Affichage graphique.
# Carte UV:
plt.close()
fig, ax = plt.subplots(1, 1)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)
ccuvusa.plot(column='uvrate', ax=ax, legend=True, cax=cax)
plt.title('UV Wh/m2', fontsize=6)
plt.savefig("uvrate.png")	

# Carte mélanome toutes ethnies:
plt.close()
fig, ax = plt.subplots(1, 1)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)
ccuvusa.plot(column='Crude Rate', ax=ax, legend=True, cax=cax)
plt.title("Melanoma per 100000", fontsize=6) 
plt.savefig("cancer.png")

# Carte mélanome blancs non hispaniques:
plt.close()
fig, ax = plt.subplots(1, 1)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)
ccuvusa.plot(column='Crude Rate W', ax=ax, legend=True, cax=cax)
plt.title("Melanoma per 100000 White", fontsize=5)
plt.savefig("cancerW.png")

# Régressions avec Koala
plt.close()
k = Koala(ccuvusa)
rall = k.reg("uvrate", "Crude Rate")
rall.fig.savefig("canceruvall.png")

rwhite = k.reg("uvrate", "Crude Rate W")
rwhite.fig.savefig("canceruvwhite.png")

print("Corrélation pour toutes ethnies (linéaire, exponentielle, puissance):")
print(rall.lin.r)
print(rall.exp.r)
print(rall.pwr.r)

print("Corrélation pour blancs non hispaniques (linéaire, exponentielle, puissance):")
print(rwhite.lin.r)
print(rwhite.exp.r)
print(rwhite.pwr.r)

# Superposition des couches graphiques UV et mélanome white
plt.close()
fig, ax = plt.subplots(1, 1)
divider = make_axes_locatable(ax)
cax = divider.append_axes("right", size="5%", pad=0.1)
base = ccuvusa.plot(column='Crude Rate W', ax=ax, legend=True, cax=cax)
plt.title("Melanoma per 100000 White", fontsize=5)

pt = ccuvusa
pt['center'] = pt.geometry.centroid
pt = pt.loc[:, ['center', 'uvrate']]
pt.columns = ['geometry', 'uvrate']

ms = np.exp(pt.uvrate / 1000) / 2

pt.plot(ax=ax, color='red', marker='o', legend=True, markersize=ms)
plt.savefig("uvcancerW.png")