3. Feature Engineering

Question 1

What is feature engineering?

Answer 1

The process of transforming raw data to useful features for model training.

Question 2

What are the primary reasons for data transformation?

Answer 2

Data compatibility, i.e., from sting type data to numerical data
Data quality, i.e., convert text to lowercase.

Question 2

What are the approaches for consistent data preprocessing?

Answer 2

Pretraining data transformation: data transformation before training. Adv: only perform once.
Disadv: update requires rerun the whole dataset.
Inside model data transformation: transformation is a part of the model code.
Adv: easy to decouple data and transformation.
Disadv: increase model latency.

Question 3

What is encoding for structured data types?

Answer 3

Categorical data must be converted to numerical as most models can’t handle categorical data directly.

Question 4

What are the two kinds of transformation may be needed for integer of floating-point data?

Answer 4

Normalization and bucketing

Question 5

Why do we need to normalize data with various ranges?

Answer 5

Slow convergence for models with gradient descent.
Wide range of values in a single feature will lead to generation of NaN error in some models.

Question 6

What are the two ways of bucketing?

Answer 6

Bucketing is to transform numeric data to categorical data.
Buckets with equal-spaced boundaries
Buckets with quantile boundaries

Question 7

What is label encoding?

Answer 7

Convert text categories to numeric while preserving the order, e.g., small, medium, big to 1, 2, 3

Question 8

What is Out of Vocab?

Answer 8

It is a special category for outliers. ML systems won’t waste time on training the rare outliers.

Question 8

What is one-hot encoding?

Answer 8

Create dummy variables used for categorical variables where order does not matter.

Question 9

What is feature hashing?

Answer 9

Apply a hash function to a categorical feature and use the hash value as the feature name.

Question 10

What is hybrid of hashing and vocabulary?

Answer 10

Use vocabulary for important features
Use hashing for the less important features

Question 11

What is embedding?

Answer 11

Embedding is a categorical feature represented as a continuous-valued feature.

Question 12

What is feature selection?

Answer 12

Select a subset of features that are most useful to a model in order to predict the target variable.

Question 13

What are the benefits of dimensionality reduction?

Answer 13

Reduce the noise from data
Reduce overfitting problem

Question 13

What are the two ways to achieve dimensionality reduction?

Answer 13

Use feature importance
Use combinations of feature, e.g., PCA, t-SNE

Question 13

What are key outcomes of classification models?

Answer 13

True positive: Predict positive class correctly
True negative: Predict negative class correctly
False positive: Predict positive class incorrectly
False negative: Predict positive class incorrectly

Question 13

What is classification threshold?

Answer 13

It is a threshold separate positive to negative class. By default, it is set at 0.5.

Question 14

What is AUC ROC used for?

Answer 14

Balanced datasets in classification problems

Question 15

What will happen if you raise and lower the classification threshold?

Answer 15

Raise: reduce false positives, increase false negatives, increase precision
Lower: reduce false negatives, increase false positives, increase recall

Question 16

What is AUC PR used for?

Answer 16

Imbalanced datasets in classification problems

Question 16

What is AUC ROC?

Answer 16

It is a graph showing the performance of a classification model at all classification thresholds.
True positive rate against False positive rate
AUC = 1, means perfect class separation

Question 17

What is AUC PR?

Answer 17

Precision values against Recall values
It gives more attention to the minor class.

Question 18

What is feature crossing?

Answer 18

Multiply two or more features.

Question 19

What are the two ways to use feature cross?

Answer 19

Cross two features: more predictive feature
Cross two or more features: represent nonlinearity

Question 20

What is TensorFlow Data API (tf.data)?

Answer 20

It is to make data input pipeline more efficient.

Question 21

What is TensorFlow Transform?

Answer 21

TensorFlow Transform library is a part of TensorFlow Extended. It performs transformations prior to training the model.
tf.Transform can avoid training-serving skew

Question 21

What is the best practice to make an efficient data input pipeline?

Answer 21

tf.data.Dataset.interleave: It parallelizes data reading.
tf.data.Dataset.cache: Cache a dataset in memory or local storage.
tf.data.Dataset.prefetch: Make sure preprocessed is ready before training
Vectorize user-defined functions on a batch of datasets.
Apply interleave, prefetch and shuffle to reduce memory usage

Question 22

What does TensorFlow Transform do?

Answer 22

You can create transform pipelines using Cloud Dataflow.
Analyze training data
Transform training data
Transform evaluation data
Produce metadata
Feed the model
Serve data

Question 23

What are the steps and libraries used in TFX pipeline?

Answer 23

Data extraction & validation: TFDV (Dataflow)
Data transformation: TF Transform (Dataflow)
Model training & tuning: tf.Estimators & tf.Keras (Vertex AI Training)
Model evaluation & validation: TF Model Analysis (Dataflow)
Model serving for prediction: TF Serving (Vertex AI Prediction)

Question 23

What are the two tools to help data transformation?

Answer 23

Data Fusion:
Code-free UI-based managed service for ETL or ELT pipelines from various sources.
Dataprep:
Code-free UI-based serverless tool for visually exploring, cleaning, preparing structured and unstructured data for analysis, reporting and machine learning at any scale.