import streamlit as st
import pandas as pd
import numpy as np
from sklearn import datasets
from sklearn.model_selection import train_test_split
from sklearn.ensemble import RandomForestClassifier as rf
from sklearn.metrics import accuracy_score
import matplotlib.pyplot as plt
from sklearn.metrics import confusion_matrix



class random_forest_st:
    def __init__(self, database, test_size=0.2):
        self.database = database
        self.test_size = test_size
        self.desc = '''
        # **Random Forest**

        Este algoritmo se construye en base al algoritmo de **Decision Tree**. Así, lo que se hace es:

        - Definir cantidad de estimadores (**Decision Tree**)
        - Cada estimador entrenarlo con una muestra del set de datos de entrenamiento, variando así la cantidad de variables y la cantidad de datos con la cual se entrenan estos estimadores.
        - Luego, para generar la predicción de algoritmo, lo que se hace es consultar a cada estimador su predicción y "**de manera democrática**" se escoge la opción más "**votada**"
        '''
        self.n_trees = 100
        self.min_samples_split = 2
        self.max_depth = 100
        self.n_feats = None
        self.stop_criterion = 'max_depth'


    def params(self):
        self.stop_criterion = st.radio('Criterio de termino:', options=['max_depth', 'min_samples_split'])
        if self.stop_criterion == 'max_depth': self.max_depth = st.slider('Valor max deph:', 1, 100, 10)
        elif self.stop_criterion == 'min_samples_split': self.min_samples_split = st.slider('Valor min_samples_split:', 2, 1000, 5)
        self.n_trees = st.slider('Cantidad de estimadores: ', 1, 100, 3)
        self.n_feats = st.slider('Fraccion de categorias para contruir los estimadores: ', 0.0, 1.0, 0.5)

    def solve(self):
        self.X, self.y = self.database.data, self.database.target
        X_train, X_test, y_train, y_test = train_test_split(self.X, self.y, test_size=self.test_size, random_state=1234)
        if self.stop_criterion == 'max_depth': self.sklearn_clf = rf(n_estimators=self.n_trees,
                                                                max_depth=self.max_depth,
                                                                max_features=self.n_feats,
                                                                random_state=1234)
        elif self.stop_criterion == 'min_samples_split': self.sklearn_clf = rf(n_estimators=self.n_trees,
                                                                min_samples_split=self.min_samples_split,
                                                                max_features=self.n_feats,
                                                                random_state=1234)

        #self.sklearn_clf = rf(n_estimators=self.n_trees)
        self.sklearn_clf.fit(X_train, y_train)
        y_pred = self.sklearn_clf.predict(X_test)
        acc = accuracy_score(y_pred, y_test)

        c1, c2 = st.columns([4, 1])
        c2.metric('Acierto', value=f'{np.round(acc, 2)*100}%')
        df = pd.DataFrame(confusion_matrix(y_pred, y_test))
        labels = self.database.target_names
        df.columns = labels
        df.index = labels
        c1.write('**Confusion Matrix**')
        c1.dataframe(df)

    def visualization(self):
        n_features = int(self.database.data.shape[1])
        self.x_feature = st.slider('Variables en eje x', 1, n_features, 1)
        self.y_feature = st.slider('Variables en eje y', 1, n_features, 2)

        self.X = np.c_[self.database.data[:, self.x_feature-1:self.x_feature], self.database.data[:, self.y_feature-1:self.y_feature]]
        self.y = self.database.target
        X_train, X_test, y_train, y_test = train_test_split(self.X, self.y, test_size=self.test_size, random_state=1234)
        if self.stop_criterion == 'max_depth': self.sklearn_clf = rf(n_estimators=self.n_trees,
                                                                max_depth=self.max_depth,
                                                                max_features=self.n_feats,
                                                                random_state=1234)
        elif self.stop_criterion == 'min_samples_split': self.sklearn_clf = rf(n_estimators=self.n_trees,
                                                                min_samples_split=self.min_samples_split,
                                                                max_features=self.n_feats,
                                                                random_state=1234)
        self.sklearn_clf.fit(X_train, y_train)

        x1_min, x1_max = self.X[:, 0].min() - 0.5, self.X[:, 0].max() + 0.5
        x2_min, x2_max = self.X[:, 1].min() - 0.5, self.X[:, 1].max() + 0.5
        h = 0.02 # Salto que vamos dando
        x1_i = np.arange(x1_min, x1_max, h)
        x2_i = np.arange(x2_min, x2_max, h)
        x1_x1, x2_x2 = np.meshgrid(x1_i, x2_i)
        y_pred = self.sklearn_clf.predict(np.c_[x1_x1.ravel(), x2_x2.ravel()])
        y_pred = y_pred.reshape(x1_x1.shape)

        plt.figure(1, figsize=(12, 8))
        plt.pcolormesh(x1_x1, x2_x2, y_pred, cmap=plt.cm.Paired)
        plt.scatter(self.X[:, 0], self.X[:, 1], c=self.y, edgecolors='k', cmap=plt.cm.Paired)
        plt.xlim(x1_x1.min(), x1_x1.max())
        plt.ylim(x2_x2.min(), x2_x2.max())
        return plt.gcf()