df=pd.read_csv('DTMNegatif.csv', index_col=0)
train, validate, test = np.split(df.sample(frac=1), [int(.6*len(df)), int(.8*len(df))])
print (train)
train.to_csv("trainNegatif.csv", sep=',')
validate.to_csv("validateNegatif.csv",sep=',')
test.to_csv("testNegatif.csv",sep=',')

#CODE TRAINING DATA FOR NAIVE BAYES
df=pd.read_csv('trainNegatif.csv', index_col=0)
df2=df.copy()
columns_names = list(df.columns.values)
total_kata=[]
list_kata=[]

for col in columns_names:
    df.loc['Total',col] =  df[col].sum()
    #sum1 = df[col].sum()
    if df[col].sum()<2:
            del df2[col]
    else:
            df2.loc['Total',col] =  df2[col].sum()
            total_kata.append(df2[col].sum())
            list_kata.append(col)

num = np.zeros(shape=(len(total_kata), 2), dtype=object)
for n, (total_kata, list_kata) in enumerate(zip(total_kata, list_kata)):
    num[n,0]=total_kata
    num[n,1]=list_kata

df3 = pd.DataFrame({'Kata':num[:,1],'Frekuensi':num[:,0]})
df2.to_csv("SeleksiFiturNegatif.csv", sep=',')
df3.to_csv("tabelFrekuensiNegatif.csv", sep=',')