import sys
import kaggle
import pandas as pd
import numpy as np
from sklearn import preprocessing
from sklearn.linear_model import LinearRegression
from sklearn.metrics import mean_squared_error
import tensorflow as tf
from tensorflow.keras.layers import Input, Dense, Activation,Dropout
from tensorflow.keras.models import Model
from tensorflow.keras.callbacks import EarlyStopping
from keras.models import Sequential

device = 'cpu'

# kaggle
kaggle.api.authenticate()
kaggle.api.dataset_download_files('timmate/avocado-prices-2020', path='.', unzip=True)

# wczytanie danych
avocado_with_year = pd.read_csv('avocado-updated-2020.csv')

# usuniecie redundantnej kolumny 'year' i zamiana wartosci 'type' na 0 lub 1
new = ['date', 'average_price', 'total_volume', '4046', '4225', '4770', 'total_bags', 'small_bags', 'large_bags', 'xlarge_bags', 'type', 'geography']
avocado = avocado_with_year[new]
avocado.to_csv("avocado.csv", index=False)
avocado = pd.read_csv('avocado.csv')
avocado['type'] = avocado.type.map(dict(organic=1, conventional=0))

# usuniecie wierszy z pustymi wartosciami
avocado.isnull().sum()
avocado.dropna()

# preprocessing
num_values = avocado.select_dtypes(include='float64').values
scaler = preprocessing.MinMaxScaler()
x_scaled = scaler.fit_transform(num_values)
num_columns = avocado.select_dtypes(include='float64').columns
avocado_normalized = pd.DataFrame(x_scaled, columns=num_columns)
for col in avocado.columns:
    if col in num_columns: 
        avocado[col] = avocado_normalized[col]
        
avocado_normalized['type'] = avocado['type']
avocado_normalized['geography'] = avocado['geography']


# podział na train/dev/test
avocado_train, avocado_validate, avocado_test = np.split(avocado_normalized.sample(frac=1), [int(.6*len(avocado_normalized)), int(.8*len(avocado_normalized))])

print("Avocado: ".ljust(20), np.size(avocado_normalized))
print("Avocado (train) : ".ljust(20), np.size(avocado_train))
print("Avocado (validate): ".ljust(20), np.size(avocado_validate))
print("Avocado (test) ".ljust(20), np.size(avocado_test))

# sprawdzenie danych
avocado_normalized.describe(include = 'all')
avocado_train.describe(include= 'all')
avocado_validate.describe(include = 'all')
avocado_test.describe(include = 'all')

avocado_normalized.geography.value_counts() 
avocado_test.geography.value_counts() 
avocado_train.geography.value_counts() 

pd.value_counts(avocado_normalized['type']).plot.bar()
pd.value_counts(avocado_train['type']).plot.bar()
pd.value_counts(avocado_test['type']).plot.bar()

avocado_normalized['average_price'].hist()
avocado_train['average_price'].hist()
avocado_validate['average_price'].hist()
avocado_test['average_price'].hist()


# print(avocado_train[:10])
# print(avocado_test[:10])
print(avocado_normalized)

# podzial na X i y
X_train = avocado_train[['average_price', 'total_volume', '4046', '4225', '4770', 'total_bags', 'small_bags', 'large_bags', 'xlarge_bags']]
y_train = avocado_train[['type']]
X_test = avocado_test[['average_price', 'total_volume', '4046', '4225', '4770', 'total_bags', 'small_bags', 'large_bags', 'xlarge_bags']]
y_test = avocado_test[['type']]

print(X_train.shape[1])
# keras model
model = Sequential()
model.add(Dense(9, input_dim = X_train.shape[1], kernel_initializer='normal', activation='relu'))
model.add(Dense(1,kernel_initializer='normal', activation='sigmoid'))

early_stop = EarlyStopping(monitor="val_loss", mode="min", verbose=1, patience=10)

# kompilacja
model.compile(loss='binary_crossentropy', optimizer='adam', metrics=['accuracy'])

# model fit
epochs = int(sys.argv[1])
batch_size = int(sys.argv[2])

model.fit(X_train, y_train, epochs=epochs, batch_size=batch_size, validation_data=(X_test, y_test))
model.save('avocado-model.h5')

# predict
predictions = model.predict(X_test)
pd.DataFrame(predictions).to_csv('prediction_results.csv')

# ewaluacja
error = mean_squared_error(y_test, predictions)
print('Error: ', error)