In [1]:

    

import tensorflow as tf
import numpy as np
from tensorflow import data
import shutil
from datetime import datetime
from mpl_toolkits.mplot3d import Axes3D
import matplotlib.pyplot as plt

from tensorflow.contrib.learn import learn_runner
from tensorflow.contrib.learn import make_export_strategy

print(tf.__version__)


    

    




/Users/khalidsalama/anaconda/lib/python3.6/importlib/_bootstrap.py:205: RuntimeWarning: compiletime version 3.5 of module 'tensorflow.python.framework.fast_tensor_util' does not match runtime version 3.6
  return f(*args, **kwds)






    




1.4.0

In [2]:

    

train_data_files = ['data/train-data.csv']
test_data_files = ['data/test-data.csv']

model_name = 'clust-model-01'

resume = False
train = True
preprocess_features = False
extend_feature_colums = False


    

In [3]:

    

HEADER = ['key', 'x1', 'x2', 'x3', 'cluster']  
HEADER_DEFAULTS = [[0], [0.0], [0.0], [0.0], ['NA']]

FEATURE_NAMES = ['x1', 'x2', 'x3']  

UNUSED_FEATURE_NAMES = list(set(HEADER) - set(FEATURE_NAMES))

print("Header: {}".format(HEADER))
print("Input Features: {}".format(FEATURE_NAMES))
print("Unused Features: {}".format(UNUSED_FEATURE_NAMES))


    

    




Header: ['key', 'x1', 'x2', 'x3', 'cluster']
Input Features: ['x1', 'x2', 'x3']
Unused Features: ['cluster', 'key']

In [4]:

    

def parse_csv_row(csv_row):
    
    columns = tf.decode_csv(csv_row, record_defaults=HEADER_DEFAULTS)
    columns = [tf.expand_dims(tensor, -1) for tensor in columns]
    features = dict(zip(HEADER, columns))
    
    for column in UNUSED_FEATURE_NAMES:
        features.pop(column)
    
    return features

def process_features(features):
    
    if preprocess_features:
        features = features
    
    return features


    

In [5]:

    

def csv_input_fn(file_names, mode=tf.estimator.ModeKeys.TRAIN, 
                 skip_header_lines=0, 
                 num_epochs=None, 
                 batch_size=200):
    
    shuffle = False
    
    print("")
    print("* data input_fn:")
    print("================")
    print("Input file(s): {}".format(file_names))
    print("Batch size: {}".format(batch_size))
    print("Epoch Count: {}".format(num_epochs))
    print("Mode: {}".format(mode))
    print("Shuffle: {}".format(shuffle))
    print("================")
    print("")

    dataset = data.TextLineDataset(filenames=file_names)
    dataset = dataset.skip(skip_header_lines)
    
    if shuffle:
        dataset = dataset.shuffle(buffer_size=2 * batch_size + 1)
    
    dataset = dataset.batch(batch_size)
    dataset = dataset.map(lambda csv_row: parse_csv_row(csv_row))
    dataset = dataset.map(lambda features: process_features(features))
    
    dataset = dataset.repeat(num_epochs)
    iterator = dataset.make_one_shot_iterator() 
    
    features = iterator.get_next()
    return features, None


    

In [6]:

    

features, _ = csv_input_fn(file_names=train_data_files)
print("Feature read from CSV: {}".format(list(features.keys())))


    

    




* data input_fn:
================
Input file(s): ['data/train-data.csv']
Batch size: 200
Epoch Count: None
Mode: train
Shuffle: False
================

Feature read from CSV: ['x1', 'x2', 'x3']

In [7]:

    

def create_estimator(run_config, hparams):
    
    estimator = tf.contrib.learn.KMeansClustering(
        num_clusters = hparams.num_clusters,
        initial_clusters= tf.contrib.factorization.RANDOM_INIT,
        distance_metric= tf.contrib.factorization.SQUARED_EUCLIDEAN_DISTANCE,
        use_mini_batch=True,
        mini_batch_steps_per_iteration=1,
        kmeans_plus_plus_num_retries=10,
        relative_tolerance=None,
        config= run_config
    )

    print("")
    print("Estimator Type: {}".format(type(estimator)))
    print("")
    
    return estimator


    

In [8]:

    

hparams  = tf.contrib.training.HParams(
    num_epochs = 1000,
    batch_size = 500,
    num_clusters=3
)

model_dir = 'trained_models/{}'.format(model_name)

run_config = tf.contrib.learn.RunConfig(
    save_checkpoints_steps=100,
    tf_random_seed=19850610,
    model_dir=model_dir
)

print(run_config.model_dir)


    

    




trained_models/clust-model-01

In [9]:

    

train_input_fn = lambda: csv_input_fn(
            train_data_files,
            mode = tf.contrib.learn.ModeKeys.TRAIN,
            num_epochs=hparams.num_epochs,
            batch_size=hparams.batch_size
        )


    

In [10]:

    

if not resume:
    print("Removing previous artifacts...")
    shutil.rmtree(model_dir, ignore_errors=True)
else:
    print("Resuming training...") 

if train:
    tf.logging.set_verbosity(tf.logging.INFO)
    
    time_start = datetime.utcnow() 
    print("Training started at {}".format(time_start.strftime("%H:%M:%S")))
    print(".......................................") 
    
    estimator = create_estimator(hparams=hparams, run_config=run_config)
    estimator.fit(input_fn=train_input_fn,
                  max_steps=None)

    time_end = datetime.utcnow() 
    print(".......................................")
    print("Training finished at {}".format(time_end.strftime("%H:%M:%S")))
    print("")
    time_elapsed = time_end - time_start
    print("Training elapsed time: {} seconds".format(time_elapsed.total_seconds()))


    

    




Removing previous artifacts...
Training started at 20:36:13
.......................................
INFO:tensorflow:Using config: {'_task_type': None, '_task_id': 0, '_cluster_spec': <tensorflow.python.training.server_lib.ClusterSpec object at 0x124bd7358>, '_master': '', '_num_ps_replicas': 0, '_num_worker_replicas': 0, '_environment': 'local', '_is_chief': True, '_evaluation_master': '', '_tf_config': gpu_options {
  per_process_gpu_memory_fraction: 1
}
, '_tf_random_seed': 19850610, '_save_summary_steps': 100, '_save_checkpoints_secs': None, '_log_step_count_steps': 100, '_session_config': None, '_save_checkpoints_steps': 100, '_keep_checkpoint_max': 5, '_keep_checkpoint_every_n_hours': 10000, '_model_dir': 'trained_models/clust-model-01'}

Estimator Type: <class 'tensorflow.contrib.learn.python.learn.estimators.kmeans.KMeansClustering'>


* data input_fn:
================
Input file(s): ['data/train-data.csv']
Batch size: 500
Epoch Count: 1000
Mode: train
Shuffle: False
================

WARNING:tensorflow:From /Users/khalidsalama/anaconda/lib/python3.6/site-packages/tensorflow/contrib/learn/python/learn/estimators/kmeans.py:120: get_global_step (from tensorflow.contrib.framework.python.ops.variables) is deprecated and will be removed in a future version.
Instructions for updating:
Please switch to tf.train.get_global_step
INFO:tensorflow:Create CheckpointSaverHook.
INFO:tensorflow:Saving checkpoints for 1 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:loss = 38770.6, step = 1
INFO:tensorflow:Saving checkpoints for 101 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 128.546
INFO:tensorflow:loss = 11287.3, step = 101 (0.779 sec)
INFO:tensorflow:Saving checkpoints for 201 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 114.686
INFO:tensorflow:loss = 10263.9, step = 201 (0.872 sec)
INFO:tensorflow:Saving checkpoints for 301 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 135.561
INFO:tensorflow:loss = 10424.8, step = 301 (0.738 sec)
INFO:tensorflow:Saving checkpoints for 401 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 124.303
INFO:tensorflow:loss = 11246.8, step = 401 (0.804 sec)
INFO:tensorflow:Saving checkpoints for 501 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 131.144
INFO:tensorflow:loss = 10249.6, step = 501 (0.763 sec)
INFO:tensorflow:Saving checkpoints for 601 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 123.435
INFO:tensorflow:loss = 10426.0, step = 601 (0.810 sec)
INFO:tensorflow:Saving checkpoints for 701 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 130.559
INFO:tensorflow:loss = 11244.1, step = 701 (0.767 sec)
INFO:tensorflow:Saving checkpoints for 801 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 137.358
INFO:tensorflow:loss = 10247.2, step = 801 (0.727 sec)
INFO:tensorflow:Saving checkpoints for 901 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 120.737
INFO:tensorflow:loss = 10427.0, step = 901 (0.828 sec)
INFO:tensorflow:Saving checkpoints for 1001 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 155.19
INFO:tensorflow:loss = 11243.2, step = 1001 (0.645 sec)
INFO:tensorflow:Saving checkpoints for 1101 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 139.163
INFO:tensorflow:loss = 10246.3, step = 1101 (0.718 sec)
INFO:tensorflow:Saving checkpoints for 1201 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 156.517
INFO:tensorflow:loss = 10427.5, step = 1201 (0.639 sec)
INFO:tensorflow:Saving checkpoints for 1301 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 135.149
INFO:tensorflow:loss = 11242.8, step = 1301 (0.740 sec)
INFO:tensorflow:Saving checkpoints for 1401 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 156.145
INFO:tensorflow:loss = 10245.8, step = 1401 (0.640 sec)
INFO:tensorflow:Saving checkpoints for 1501 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 134.041
INFO:tensorflow:loss = 10427.9, step = 1501 (0.746 sec)
INFO:tensorflow:Saving checkpoints for 1601 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 155.929
INFO:tensorflow:loss = 11242.5, step = 1601 (0.641 sec)
INFO:tensorflow:Saving checkpoints for 1701 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 155.166
INFO:tensorflow:loss = 10245.5, step = 1701 (0.644 sec)
INFO:tensorflow:Saving checkpoints for 1801 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 134.543
INFO:tensorflow:loss = 10428.1, step = 1801 (0.743 sec)
INFO:tensorflow:Saving checkpoints for 1901 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 155.267
INFO:tensorflow:loss = 11242.4, step = 1901 (0.644 sec)
INFO:tensorflow:Saving checkpoints for 2001 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 138.506
INFO:tensorflow:loss = 10245.3, step = 2001 (0.722 sec)
INFO:tensorflow:Saving checkpoints for 2101 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 158.153
INFO:tensorflow:loss = 10428.3, step = 2101 (0.633 sec)
INFO:tensorflow:Saving checkpoints for 2201 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 135.475
INFO:tensorflow:loss = 11242.3, step = 2201 (0.738 sec)
INFO:tensorflow:Saving checkpoints for 2301 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 156.354
INFO:tensorflow:loss = 10245.2, step = 2301 (0.640 sec)
INFO:tensorflow:Saving checkpoints for 2401 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 151.1
INFO:tensorflow:loss = 10428.5, step = 2401 (0.662 sec)
INFO:tensorflow:Saving checkpoints for 2501 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 134.795
INFO:tensorflow:loss = 11242.2, step = 2501 (0.742 sec)
INFO:tensorflow:Saving checkpoints for 2601 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 153.641
INFO:tensorflow:loss = 10245.1, step = 2601 (0.651 sec)
INFO:tensorflow:Saving checkpoints for 2701 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 135.826
INFO:tensorflow:loss = 10428.6, step = 2701 (0.736 sec)
INFO:tensorflow:Saving checkpoints for 2801 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 158.035
INFO:tensorflow:loss = 11242.1, step = 2801 (0.633 sec)
INFO:tensorflow:Saving checkpoints for 2901 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:global_step/sec: 134.376
INFO:tensorflow:loss = 10245.0, step = 2901 (0.744 sec)
INFO:tensorflow:Saving checkpoints for 3000 into trained_models/clust-model-01/model.ckpt.
INFO:tensorflow:Loss for final step: 11242.1.
.......................................
Training finished at 20:36:36

Training elapsed time: 23.348445 seconds

In [11]:

    

train_input_fn = lambda: csv_input_fn(
            train_data_files,
            num_epochs=1,
            batch_size=1500
        )

test_input_fn = lambda: csv_input_fn(
            test_data_files,
            num_epochs=1,
            batch_size=500
        )

train_assignments = list(estimator.predict_cluster_idx(input_fn=train_input_fn))
test_assignments = list(estimator.predict_cluster_idx(input_fn=test_input_fn))


    

    




* data input_fn:
================
Input file(s): ['data/train-data.csv']
Batch size: 1500
Epoch Count: 1
Mode: train
Shuffle: False
================

INFO:tensorflow:Restoring parameters from trained_models/clust-model-01/model.ckpt-3000

* data input_fn:
================
Input file(s): ['data/test-data.csv']
Batch size: 500
Epoch Count: 1
Mode: train
Shuffle: False
================

INFO:tensorflow:Restoring parameters from trained_models/clust-model-01/model.ckpt-3000

In [12]:

    

import pandas as pd
import numpy as np

train_df = pd.read_csv(train_data_files[0], header=None, index_col=0)
test_df = pd.read_csv(test_data_files[0], header=None, index_col=0)


fig = plt.figure(figsize=(20,10))
ax = fig.add_subplot(121, projection='3d')

ax.scatter(train_df.iloc[:,0], train_df.iloc[:,1], train_df.iloc[:,2], c=train_assignments, marker='o')

ax = fig.add_subplot(122, projection='3d')

ax.scatter(test_df.iloc[:,0], test_df.iloc[:,1], test_df.iloc[:,2], c=test_assignments, marker='o')
plt.show()


    

In [13]:

    

clusters = estimator.clusters()
print("Cluster Centriods:")
print("==================")
print(clusters)


    

    




Cluster Centriods:
==================
[[ 8.26997852  0.80968368  3.60038066]
 [-0.66113287 -8.35648823 -5.18023205]
 [-5.76904392  4.08977938 -8.63212681]]

In [ ]: