import pandas as pd
import numpy as np


data = pd.read_csv('car.data',header=None)

X = data.iloc[:,:-1].values
Y = data.iloc[:,-1].values
    
class Bayes:
    def __init__(self):
        self.prob_Y = {}
        self.prob_con = {}
        self.gaussians = {}
        
    def fit(self,X,Y):
        klasy =np.unique(Y)
        self.klasy = klasy
        self.prob_Y = {}
        for klasa in klasy:
            self.prob_Y[klasa] = X[klasa == Y]/len(X)

        #prawdopodobienstwo warunkowe
        for kl in klasy:
            lista_dla_klasy = []
            for at in range(X.shape[1]):
                uniques = np.unique(X[Y == kl][:,at])
                sl = {}
                for u in uniques:
                    sl[u] = sum((X[Y == kl][:,at] == u)/len(X[Y == kl]))
                lista_dla_klasy.append(sl)
            self.prob_con[kl] = lista_dla_klasy    
        
        #gauss
        for kl in klasy:
            means = np.mean(X[Y==kl],axis = 0)
            stds = np.std(X[Y==kl],axis = 0)
            self.gaussians[kl] = [(means[i],stds[i]) for i in range(len(means))]

    def predict(self, tab_X):
        y_hat = []
        for x in tab_X:
            wskazniki_dla_klas = np.array(np.array(list(self.prob_Y.items()))[:,1], dtype=float)
            for klasa in range(len(self.klasy)):
                for idw, wartosc in enumerate(x):
                    if isinstance(wartosc,(int,float)):
                        mean,std = self.gaussians[self.klasy[klasa]][idw]
                        wskazniki_dla_klas[klasa] *= (1 / (np.sqrt(2 * np.pi) * std)) * np.exp(-((wartosc - mean) ** 2) / (2 * std ** 2))
                    else:
                        wskazniki_dla_klas[klasa] *= self.prob_con[self.klasy[klasa]][idw][wartosc]
                    
            y_hat.append(self.klasy[np.argmax(wskazniki_dla_klas)])
        return y_hat
 
    
    def evaluate(self,X,Y):
        return sum(self.predict(X) == Y)/len(Y)