先試後買

A. They are usually referred to as "black box" models.

B. They are usually binary classifiers.

C. They usually compromise on model accuracy for the sake of interpretability.

D. They are usually easier to explain to business stakeholders.

E. They are usually very good at solving non-linear problems.

A. K-fold

B. Leave-one-out

C. Bootstrapping

D. Stratification

A. Unsupervised Learning

B. Dijkstra Algorithm

C. Reinforcement learning

D. Supervised Learning.

A. Words are featurized by taking a matrix of bigram counts.

B. Words are featurized by taking a histogram of letter counts.

C. Words are grouped together into clusters and then represented by word cluster membership.

D. Words are converted into numerical vectors.

A. De-Duplicate

B. Duplicate

C. Destroy

D. Detain

A. Platform as a Service (PaaS) can provide some services within an application such as payment applications to create efficient results.

B. Software as a Service (SaaS) can provide AI practitioner data science services such as Jupyter notebooks.

C. Infrastructure as a Service (IaaS) can provide CPU, memory, disk, network and GPU.

D. Data as a Service (DaaS) can host the databases providing backups, clustering, and high availability.

A. Change the way output is visualized in the monitoring step.

B. Change in normalization function in the feature engineering step.

C. Add a new method for drift detection in the model evaluation step.

D. Add a new pipeline for retraining the model in the model training step.

A. Increases recall over precision

B. Makes models easier to explain to business stakeholders

C. Improves model interpretability

D. Controls the learning process of a given algorithm

A. Avoiding mechanisms to explain and justify automated decisions.

B. Understanding, documenting, and displaying data lineage.

C. Utilizing quasi-identifiers and non-unique identifiers, alone or in combination.

D. Collecting and processing the largest amount of data possible.

A. Pandas

B. IBM Watson

C. Spark ML

D. Databricks