Notice: This page requires JavaScript to function properly.
Please enable JavaScript in your browser settings or update your browser.Data Preprocessing | Basics of Keras
Neural Networks with TensorFlow
course content

Course Content

Neural Networks with TensorFlow

Neural Networks with TensorFlow

1. Basics of Keras
LayersModel CreationModel CompilationData PreprocessingModel Training and EvaluationChallengeModel Save and LoadEarly Stopping and CheckpointsHyperparameter TuningChallengeQuiz
2. Regularization
Overfitting and UnderfittingWhat is Regularization?L1, L2 RegularizationDropoutBatch NormalizationChallengeQuiz
3. Advanced Techniques
OptimizersLearning Rate SchedulingTensorFlow DatasetsData GeneratorsNon-Sequential ModelsTransfer LearningMultitask LearningChallengeQuizSummary

bookData Preprocessing

Now, let's explore setting up a model and apply it to a practical scenario. We'll aim to predict house prices using the well-known Boston Housing Price Regression Dataset.

Data Overview

First, we need to examine the data before loading it.

Note

For an in-depth description of the dataset, you can visit this link: Boston Housing price regression dataset.

Features

Below is a list of all the columns in the dataset:

Ethical Consideration: The dataset includes a variable "B," which implies a correlation between race and house prices. We will exclude this column in the preprocessing stage to avoid racial bias in our model.

Note

To understand more about ethical considerations, check out our Ethical Considerations in Deep Learning article.

Missing Values

We need to verify if there are any missing values in the dataset. This requires first loading the dataset from Keras and then evaluating for missing values.


              
12345678910111213

            
import tensorflow.keras as keras
import pandas as pd

# Load the dataset
(X_train, y_train), (X_test, y_test) = keras.datasets.boston_housing.load_data()

# Convert each subset to Pandas DataFrame
X_train = pd.DataFrame(X_train)
y_train = pd.DataFrame(y_train)

# Sum up number of empty values of each set
print('Null values in X_Train:', X_train.isnull().sum().sum())
print('Null values in y_train:', y_train.isnull().sum().sum())
copy

As it turns out, there are no empty values, so we don't need to address this issue.

Data Preprocessing

  • Outliers: Although Keras datasets are typically free of outliers, we will demonstrate outlier removal using IsolationForest, eliminating 5% of the data as outliers.

    Note

    • IsolationForest's predict method returns a list indicating valid samples (1) or outliers (-1).
    • To set up contamination rate you can set up contamination parameter of the IsolationForest constructor.
    • Outliers should be removed only from the training set.
  • Rescaling: To ensure consistency and compatibility with our model, the data needs to be rescaled.

Task

Throughout this course, you'll engage extensively in data handling independently. Therefore, let's revisit how to preprocess data using the scikit-learn library, making it ready for use in a neural network developed with Keras.

  1. Initialize an Isolation Forest with a 5% contamination rate.
  2. Apply the Isolation Forest to the training set and discard outliers.
  3. Rescale the data using MinMaxScaler.

Switch to desktopSwitch to desktop for real-world practiceContinue from where you are using one of the options below
Everything was clear?

How can we improve it?

Thanks for your feedback!

Section 1. Chapter 4
toggle bottom row

bookData Preprocessing

Now, let's explore setting up a model and apply it to a practical scenario. We'll aim to predict house prices using the well-known Boston Housing Price Regression Dataset.

Data Overview

First, we need to examine the data before loading it.

Note

For an in-depth description of the dataset, you can visit this link: Boston Housing price regression dataset.

Features

Below is a list of all the columns in the dataset:

Ethical Consideration: The dataset includes a variable "B," which implies a correlation between race and house prices. We will exclude this column in the preprocessing stage to avoid racial bias in our model.

Note

To understand more about ethical considerations, check out our Ethical Considerations in Deep Learning article.

Missing Values

We need to verify if there are any missing values in the dataset. This requires first loading the dataset from Keras and then evaluating for missing values.


              
12345678910111213

            
import tensorflow.keras as keras
import pandas as pd

# Load the dataset
(X_train, y_train), (X_test, y_test) = keras.datasets.boston_housing.load_data()

# Convert each subset to Pandas DataFrame
X_train = pd.DataFrame(X_train)
y_train = pd.DataFrame(y_train)

# Sum up number of empty values of each set
print('Null values in X_Train:', X_train.isnull().sum().sum())
print('Null values in y_train:', y_train.isnull().sum().sum())
copy

As it turns out, there are no empty values, so we don't need to address this issue.

Data Preprocessing

  • Outliers: Although Keras datasets are typically free of outliers, we will demonstrate outlier removal using IsolationForest, eliminating 5% of the data as outliers.

    Note

    • IsolationForest's predict method returns a list indicating valid samples (1) or outliers (-1).
    • To set up contamination rate you can set up contamination parameter of the IsolationForest constructor.
    • Outliers should be removed only from the training set.
  • Rescaling: To ensure consistency and compatibility with our model, the data needs to be rescaled.

Task

Throughout this course, you'll engage extensively in data handling independently. Therefore, let's revisit how to preprocess data using the scikit-learn library, making it ready for use in a neural network developed with Keras.

  1. Initialize an Isolation Forest with a 5% contamination rate.
  2. Apply the Isolation Forest to the training set and discard outliers.
  3. Rescale the data using MinMaxScaler.

Switch to desktopSwitch to desktop for real-world practiceContinue from where you are using one of the options below
Everything was clear?

How can we improve it?

Thanks for your feedback!

Section 1. Chapter 4
toggle bottom row

bookData Preprocessing

Now, let's explore setting up a model and apply it to a practical scenario. We'll aim to predict house prices using the well-known Boston Housing Price Regression Dataset.

Data Overview

First, we need to examine the data before loading it.

Note

For an in-depth description of the dataset, you can visit this link: Boston Housing price regression dataset.

Features

Below is a list of all the columns in the dataset:

Ethical Consideration: The dataset includes a variable "B," which implies a correlation between race and house prices. We will exclude this column in the preprocessing stage to avoid racial bias in our model.

Note

To understand more about ethical considerations, check out our Ethical Considerations in Deep Learning article.

Missing Values

We need to verify if there are any missing values in the dataset. This requires first loading the dataset from Keras and then evaluating for missing values.


              
12345678910111213

            
import tensorflow.keras as keras
import pandas as pd

# Load the dataset
(X_train, y_train), (X_test, y_test) = keras.datasets.boston_housing.load_data()

# Convert each subset to Pandas DataFrame
X_train = pd.DataFrame(X_train)
y_train = pd.DataFrame(y_train)

# Sum up number of empty values of each set
print('Null values in X_Train:', X_train.isnull().sum().sum())
print('Null values in y_train:', y_train.isnull().sum().sum())
copy

As it turns out, there are no empty values, so we don't need to address this issue.

Data Preprocessing

  • Outliers: Although Keras datasets are typically free of outliers, we will demonstrate outlier removal using IsolationForest, eliminating 5% of the data as outliers.

    Note

    • IsolationForest's predict method returns a list indicating valid samples (1) or outliers (-1).
    • To set up contamination rate you can set up contamination parameter of the IsolationForest constructor.
    • Outliers should be removed only from the training set.
  • Rescaling: To ensure consistency and compatibility with our model, the data needs to be rescaled.

Task

Throughout this course, you'll engage extensively in data handling independently. Therefore, let's revisit how to preprocess data using the scikit-learn library, making it ready for use in a neural network developed with Keras.

  1. Initialize an Isolation Forest with a 5% contamination rate.
  2. Apply the Isolation Forest to the training set and discard outliers.
  3. Rescale the data using MinMaxScaler.

Switch to desktopSwitch to desktop for real-world practiceContinue from where you are using one of the options below
Everything was clear?

How can we improve it?

Thanks for your feedback!

Now, let's explore setting up a model and apply it to a practical scenario. We'll aim to predict house prices using the well-known Boston Housing Price Regression Dataset.

Data Overview

First, we need to examine the data before loading it.

Note

For an in-depth description of the dataset, you can visit this link: Boston Housing price regression dataset.

Features

Below is a list of all the columns in the dataset:

Ethical Consideration: The dataset includes a variable "B," which implies a correlation between race and house prices. We will exclude this column in the preprocessing stage to avoid racial bias in our model.

Note

To understand more about ethical considerations, check out our Ethical Considerations in Deep Learning article.

Missing Values

We need to verify if there are any missing values in the dataset. This requires first loading the dataset from Keras and then evaluating for missing values.


              
12345678910111213

            
import tensorflow.keras as keras
import pandas as pd

# Load the dataset
(X_train, y_train), (X_test, y_test) = keras.datasets.boston_housing.load_data()

# Convert each subset to Pandas DataFrame
X_train = pd.DataFrame(X_train)
y_train = pd.DataFrame(y_train)

# Sum up number of empty values of each set
print('Null values in X_Train:', X_train.isnull().sum().sum())
print('Null values in y_train:', y_train.isnull().sum().sum())
copy

As it turns out, there are no empty values, so we don't need to address this issue.

Data Preprocessing

  • Outliers: Although Keras datasets are typically free of outliers, we will demonstrate outlier removal using IsolationForest, eliminating 5% of the data as outliers.

    Note

    • IsolationForest's predict method returns a list indicating valid samples (1) or outliers (-1).
    • To set up contamination rate you can set up contamination parameter of the IsolationForest constructor.
    • Outliers should be removed only from the training set.
  • Rescaling: To ensure consistency and compatibility with our model, the data needs to be rescaled.

Task

Throughout this course, you'll engage extensively in data handling independently. Therefore, let's revisit how to preprocess data using the scikit-learn library, making it ready for use in a neural network developed with Keras.

  1. Initialize an Isolation Forest with a 5% contamination rate.
  2. Apply the Isolation Forest to the training set and discard outliers.
  3. Rescale the data using MinMaxScaler.

Switch to desktopSwitch to desktop for real-world practiceContinue from where you are using one of the options below
Section 1. Chapter 4
Switch to desktopSwitch to desktop for real-world practiceContinue from where you are using one of the options below
some-alt