# -*- coding: utf-8 -*-
"""
Created on Sat Mar  7 10:41:51 2020

@author: Johan Rene van Dorp
"""

import pandas as pd
import numpy as np
import matplotlib.pyplot as plt
import Dist_Library as dl
from sklearn.neighbors import KernelDensity

df = pd.read_csv('OldFaithFul.csv')
Durations = df["Duration (s)"]
WaitingTimes = df["Waiting Time (Min)"]

# Plotting a two panel histogram plot of the durations
# the waiting times

plt.rc('font', family='serif', size='10')
plt.figure(figsize=(10, 5))
Hist_Figure = plt.figure()
Hist_Figure.subplots_adjust(hspace=0.4, wspace=0.4)
off_set_x = 10

# Calculating the histogram data for the left panel
Number_of_bins = 22
LB = 90
UB = 330
bins = np.linspace(LB, UB, Number_of_bins+1) 
Duration_hist_data = dl.Estimate_empirical_histogram_table(bins,Durations)

# Adding the left top panel with the theoretical pmf
the_bounds = np.append(Duration_hist_data["LB"][0],Duration_hist_data["UB"])
the_density = np.append(0,Duration_hist_data["Bin PDF"])
x_lims = (LB-off_set_x,UB+off_set_x)
y_lims = (0,0.014)
Panel = Hist_Figure.add_subplot(1,2,1)
dl.add_pmf_density_hist_to_figure_panel(Panel,the_bounds,the_density,'indianred')
plt.xlim(x_lims)
plt.ylim(y_lims)
plt.xlabel('Duration (in Sec)')
plt.ylabel('Density')
text_str = '# Bins '+str(Number_of_bins)
plt.text(160,0.013,text_str,color = 'red',size = 10)
# Adding a Kernel Density to the plot
Dur = Durations.to_numpy()
kde = KernelDensity(bandwidth=9.0, kernel='gaussian')
kde.fit(Dur[:, None])
x_d = np.linspace(LB-off_set_x, UB+off_set_x, 1000)
logprob = kde.score_samples(x_d[:, None])
plt.plot(x_d, np.exp(logprob), alpha=1, color = 'navy')

# Calculating the histogram data for the right panel
Number_of_bins = 22
LB = 40
UB = 100
bins = np.linspace(LB, UB, Number_of_bins+1) 
WaitingTime_hist_data = dl.Estimate_empirical_histogram_table(bins,WaitingTimes)

# Adding the righ top panel with the theoretical pmf
the_bounds = np.append(WaitingTime_hist_data["LB"][0],WaitingTime_hist_data["UB"])
the_density = np.append(0,WaitingTime_hist_data["Bin PDF"])
x_lims = (LB-off_set_x,UB+off_set_x)
y_lims = (0,0.07)
Panel = Hist_Figure.add_subplot(1,2,2)
dl.add_pmf_density_hist_to_figure_panel(Panel,the_bounds,the_density,'skyblue')
plt.xlim(x_lims)
plt.ylim(y_lims)
plt.xlabel('Waiting Time (in Min)')
plt.ylabel('Density')
text_str = '# Bins '+str(Number_of_bins)
plt.text(60,0.065,text_str,color = 'red',size = 10)
# Adding a Kernel Density to the plot
# Adding a Kernel Density to the plot
Wai = WaitingTimes.to_numpy()
kde = KernelDensity(bandwidth=3.0, kernel='gaussian')
kde.fit(Wai[:, None])
x_d = np.linspace(LB-off_set_x, UB+off_set_x, 1000)
logprob = kde.score_samples(x_d[:, None])
plt.plot(x_d, np.exp(logprob), alpha=1, color = 'navy')

Hist_Figure.suptitle('Histograms Old Faithful Data',size='14')
plt.savefig('Old_Faithful.png', dpi=1200)