Create your first high level TensorFlow Model



In [0]:

    
!pip install -q tf-nightly-gpu-2.0-preview



In [2]:

    
import tensorflow as tf
print(tf.__version__)









    



2.0.0-dev20190512



In [0]:

    
import matplotlib.pyplot as plt
import pandas as pd
import tensorflow as tf
import numpy as np
from tensorflow import keras



In [4]:

    
!curl -O https://raw.githubusercontent.com/DJCordhose/deep-learning-crash-course-notebooks/master/data/insurance-customers-1500.csv









    



  % Total    % Received % Xferd  Average Speed   Time    Time     Time  Current
                                 Dload  Upload   Total   Spent    Left  Speed
100 26783  100 26783    0     0   196k      0 --:--:-- --:--:-- --:--:--  196k



In [5]:

    
df = pd.read_csv('./insurance-customers-1500.csv', sep=';')
y = df['group']
df.drop('group', axis='columns', inplace=True)
X = df.as_matrix()









    



/usr/local/lib/python3.6/dist-packages/ipykernel_launcher.py:4: FutureWarning: Method .as_matrix will be removed in a future version. Use .values instead.
  after removing the cwd from sys.path.

Input data now is in X, labels are in y



In [6]:

    
df.head()



In [7]:

    
X.shape









    Out[7]:





(1500, 3)



In [8]:

    
y.head()









    Out[8]:





0    1
1    1
2    0
3    2
4    1
Name: group, dtype: int64



In [9]:

    
y.shape









    Out[9]:





(1500,)

Create your model here



In [0]:

    
from tensorflow.keras.layers import Dense



In [0]:

    
# getting help
Dense?



In [12]:

    
# this is broken, fix

model = keras.Sequential()

model.add(Dense(name='hidden1', units=??, activation=??, input_dim=3))

# how many hidden layers?
model.add(Dense(name='output', units=??, activation=??))
# model.add(Dense(name='output', units=??, activation=??))

model.summary()









    



Model: "sequential"
_________________________________________________________________
Layer (type)                 Output Shape              Param #   
=================================================================
hidden1 (Dense)              (None, 50)                200       
_________________________________________________________________
output (Dense)               (None, 3)                 153       
=================================================================
Total params: 353
Trainable params: 353
Non-trainable params: 0
_________________________________________________________________



In [0]:

    
model.fit?



In [17]:

    
%%time 

model.compile(loss='sparse_categorical_crossentropy',
             optimizer='adam',
             metrics=['accuracy'])

EPOCHS = 500

history = model.fit(X, y, epochs=EPOCHS, batch_size=2000, verbose=0)









    



CPU times: user 2.04 s, sys: 140 ms, total: 2.18 s
Wall time: 1.83 s



In [18]:

    
train_loss, train_accuracy = model.evaluate(X, y)
train_accuracy









    



1500/1500 [==============================] - 0s 70us/sample - loss: 0.7322 - accuracy: 0.6793






    Out[18]:





0.6793333



In [19]:

    
# plt.yscale('log')
plt.ylabel("accuracy")
plt.xlabel("epochs")

plt.plot(history.history['accuracy'])









    Out[19]:





[<matplotlib.lines.Line2D at 0x7f4a017066a0>]



In [20]:

    
model.predict(np.array([[100, 48, 10]]))









    Out[20]:





array([[0.26641512, 0.60835946, 0.12522537]], dtype=float32)



In [0]:

	speed	age	miles
0	98.0	44.0	25.0
1	118.0	54.0	24.0
2	111.0	26.0	34.0
3	97.0	25.0	10.0
4	114.0	38.0	22.0