import matplotlib.pyplot as plt
import numpy as np
import tensorflow as tf
import pandas as pd
import seaborn as sns
from keras.layers import Normalization, Dense, InputLayer
from keras.losses import MeanSquaredError, Huber, MeanAbsoluteError
from keras.optimizers import Adam
from keras.metrics import RootMeanSquaredError

data = pd.read_csv("/Users/atatekeli/PycharmProjects/CompVision/comp-vision-projects/tensorflow-comp-vision/Car Price Prediction/carprice.csv")
print(data.head())
print(data.shape)
print(data.describe())

sns.pairplot(data[['years', 'km', 'rating', 'condition', 'economy', 'top speed', 'hp', 'torque', 'current price']], diag_kind='kde')

tensor_data = tf.constant(data)
tensor_data = tf.cast(tensor_data, tf.float32)
print(tensor_data)

tensor_data = tf.random.shuffle(tensor_data)
print(tensor_data[:5])

X = tensor_data[:,3:-1]
print(X[:5])

y = tensor_data[:,-1]
print(y[:5].shape)
y = tf.expand_dims(y, axis = -1)
print(y[:5])

normalizer = Normalization(axis = -1, mean = 5, variance = 4)
x_normalized = tf.constant([[3,4,5,6,7],
                            [4,5,6,7,8]])
normalizer(x_normalized)

normalizer = Normalization()
x_normalized = tf.constant([[3,4,5,6,7],
                            [4,10,6,7,8],
                            [32,1,56,3,5]])
normalizer.adapt(x_normalized)
normalizer(x_normalized)

print(X.shape)

TRAIN_RATIO = 0.8
VAL_RATIO = 0.1
TEST_RATIO = 0.1
DATASET_SIZE = len(X)

X_train = X[:int(DATASET_SIZE*TRAIN_RATIO)]
y_train = y[:int(DATASET_SIZE*TRAIN_RATIO)]
print(X_train.shape)
print(y_train.shape)

train_dataset = tf.data.Dataset.from_tensor_slices((X_train, y_train))
train_dataset = train_dataset.shuffle(buffer_size = 8, reshuffle_each_iteration = True).batch(32).prefetch(tf.data.AUTOTUNE)

for x,y in train_dataset:
  print(x,y)
  break

X_val = X[int(DATASET_SIZE*TRAIN_RATIO):int(DATASET_SIZE*(TRAIN_RATIO+VAL_RATIO))]
y_val = y[int(DATASET_SIZE*TRAIN_RATIO):int(DATASET_SIZE*(TRAIN_RATIO+VAL_RATIO))]
print(X_val.shape)
print(y_val.shape)

val_dataset = tf.data.Dataset.from_tensor_slices((X_val, y_val))
val_dataset = train_dataset.shuffle(buffer_size = 8, reshuffle_each_iteration = True).batch(32).prefetch(tf.data.AUTOTUNE)

X_test = X[int(DATASET_SIZE*(TRAIN_RATIO+VAL_RATIO)):]
y_test = y[int(DATASET_SIZE*(TRAIN_RATIO+VAL_RATIO)):]
print(X_test.shape)
print(y_test.shape)

test_dataset = tf.data.Dataset.from_tensor_slices((X_test, y_test))
test_dataset = train_dataset.shuffle(buffer_size = 8, reshuffle_each_iteration = True).batch(32).prefetch(tf.data.AUTOTUNE)

normalizer = Normalization()
normalizer.adapt(X_train)
print(normalizer(X)[:5])

print(X[:5])

"""## **Model Creation and Training**"""

model = tf.keras.Sequential([
                             InputLayer(input_shape = (8,)),
                             normalizer,
                             Dense(128, activation = "relu"),
                             Dense(128, activation = "relu"),
                             Dense(128, activation = "relu"),
                             Dense(1),
])
print(model.summary())

tf.keras.utils.plot_model(model, to_file = "model.png", show_shapes=True)

model.compile(optimizer = Adam(learning_rate = 0.1),
              loss = MeanAbsoluteError(),
              metrics = RootMeanSquaredError())

history = model.fit(train_dataset, validation_data=val_dataset, epochs = 100, verbose = 1)

plt.plot(history.history['loss'])
plt.plot(history.history['val_loss'])
plt.title('model loss')
plt.ylabel('loss')
plt.xlabel('epoch')
plt.legend(['train', 'val_loss'])
plt.show()

plt.plot(history.history['root_mean_squared_error'])
plt.plot(history.history['val_root_mean_squared_error'])
plt.title('model performance')
plt.ylabel('rmse')
plt.xlabel('epoch')
plt.legend(['train', 'val'])
plt.show()

"""## **Model Evaluation and Testing**"""

model.evaluate(X_test,y_test)
print(X_test.shape)

model.predict(tf.expand_dims(X_test[0], axis = 0 ))

print(y_test[0])
y_true = list(y_test[:,0].numpy())

y_pred = list(model.predict(X_test)[:,0])
print(y_pred)

ind = np.arange(100)
plt.figure(figsize=(40,20))

width = 0.1

plt.bar(ind, y_pred, width, label='Predicted Car Price')
plt.bar(ind + width, y_true, width, label='Actual Car Price')

plt.xlabel('Actual vs Predicted Prices')
plt.ylabel('Car Price Prices')

plt.show()

data = pd.read_csv("train.csv", ",")
data.head()