orc.ipynb

Kernel: Python 3

Copyright 2021 The TensorFlow Authors.

#@title Licensed under the Apache License, Version 2.0 (the "License");
# you may not use this file except in compliance with the License.
# You may obtain a copy of the License at
#
# https://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing, software
# distributed under the License is distributed on an "AS IS" BASIS,
# WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
# See the License for the specific language governing permissions and
# limitations under the License.

Apache ORC Reader

View on TensorFlow.org

Run in Google Colab

View on GitHub

Download notebook

Overview

Apache ORC is a popular columnar storage format. tensorflow-io package provides a default implementation of reading Apache ORC files.

Setup

Install required packages, and restart runtime

!pip install tensorflow-io

import tensorflow as tf
import tensorflow_io as tfio

Download a sample dataset file in ORC

The dataset you will use here is the Iris Data Set from UCI. The data set contains 3 classes of 50 instances each, where each class refers to a type of iris plant. It has 4 attributes: (1) sepal length, (2) sepal width, (3) petal length, (4) petal width, and the last column contains the class label.

!curl -OL https://github.com/tensorflow/io/raw/master/tests/test_orc/iris.orc
!ls -l iris.orc

Create a dataset from the file

dataset = tfio.IODataset.from_orc("iris.orc", capacity=15).batch(1)

Examine the dataset:

for item in dataset.take(1):
    print(item)

Let's walk through an end-to-end example of tf.keras model training with ORC dataset based on iris dataset.

Data preprocessing

Configure which columns are features, and which column is label:

feature_cols = ["sepal_length", "sepal_width", "petal_length", "petal_width"]
label_cols = ["species"]

# select feature columns
feature_dataset = tfio.IODataset.from_orc("iris.orc", columns=feature_cols)
# select label columns
label_dataset = tfio.IODataset.from_orc("iris.orc", columns=label_cols)

A util function to map species to float numbers for model training:

vocab_init = tf.lookup.KeyValueTensorInitializer(
    keys=tf.constant(["virginica", "versicolor", "setosa"]),
    values=tf.constant([0, 1, 2], dtype=tf.int64))
vocab_table = tf.lookup.StaticVocabularyTable(
    vocab_init,
    num_oov_buckets=4)

label_dataset = label_dataset.map(vocab_table.lookup)
dataset = tf.data.Dataset.zip((feature_dataset, label_dataset))
dataset = dataset.batch(1)

def pack_features_vector(features, labels):
    """Pack the features into a single array."""
    features = tf.stack(list(features), axis=1)
    return features, labels

dataset = dataset.map(pack_features_vector)

Build, compile and train the model

Finally, you are ready to build the model and train it! You will build a 3 layer keras model to predict the class of the iris plant from the dataset you just processed.

model = tf.keras.Sequential(
    [
        tf.keras.layers.Dense(
            10, activation=tf.nn.relu, input_shape=(4,)
        ),
        tf.keras.layers.Dense(10, activation=tf.nn.relu),
        tf.keras.layers.Dense(3),
    ]
)

model.compile(optimizer="adam", loss=tf.keras.losses.SparseCategoricalCrossentropy(from_logits=True), metrics=["accuracy"])
model.fit(dataset, epochs=5)