Predictive Intelligence Flashcards
What is Predictive Intelligence?
an interface by which you can train models using the ServiceNow AI Platform.
Provide input/output pairs, so the computer can generate a rule that maps between the two. The output rule is often referred to as a model or solution.
Supervised Learning/Training
Provide raw data, allow the computer to identify structure within the input data set. The output is referred to as a model or solution.
Unsupervised Learning/Training
What attempts to explain the behavior of a target attribute, as a function of a set of independent input attributes?
Classification
What attempts to predict a numeric value for a target, as a function of a set of independent input attributes?
Regression
What is clustering?
Clustering groups similar records into into clusters so you can address them collectively or identify patterns.
What is similarity?
Given a lookup set of raw data, predict similarity (output) between a given set (input) and the raw data.
A system designed to perform tasks, usually requiring a level of human intelligence.
Artificial Intelligence (AI)
A subset of AI that provides the ability to improve from experience. This action is done by using algorithms, mathematics, and statistics to create models that make predictions/decisions.
Machine Learning (ML)
What roles are associated with Predictive Intelligence?
[ml_admin] & [ml_report_user]
The predictive model that is generated as the output of training (supervised or unsupervised) using the training data set (output of solution definition).
Solution
Given a set of correlated inputs/outputs, the solution generated will also predict an output based on user input.
Classification
Given raw data (lookup set), the solution generated will predict similarity between a given input (example set) and the raw data
Similarity
A collection of all the words and phrases that the machine can use for that particular solution to compare rows of data based on textual similarity.
Word Corpus
Given raw data, the solution generated will generate a structure across the input data set. Each set of similar records is grouped to address them collectively or identify patterns that were not detected before.
Cluster
The proportion of positive identifications that is actually correct. Calculated as the number of true positives divided by the sum of true positives and false positives.
Precision
The aggregate percentage of records that receive a prediction.
Coverage
To predict field values, which type of solution should you create?
Classification
To group similar records into clusters so you can address them collectively or identify patterns, which type of solution should be created?
Clustering
Predictive Intelligence offers out-of-the-box solutions for which applications?
ITSM, CSM, HR, and Event Management
At a minimum, which role is required to manage Predictive Intelligence solutions?
[ml_admin]
T | F: Regression is a machine-learning framework that can be trained with historical data to predict numeric outputs, such as a temperature or a stock price.
True
To surface similar tasks and content to predict new major issues and recommend critical actions, which type of solution should you create?
Similarity
When implementing PI, how many historical records does ServiceNow recommend you have?
Between 30,000 and 300,000.
This setting groups together points that are close to each other based on a distance measurement and a minimum number of points. It also marks as outliers the point that are in low-density regions.
DBSCAN
T | F: DBSCAN is only available for similarity solutions.
False: It is available for Clustering solutions only.
This setting evaluates how ‘important’ a word is to a document in the document set or word corpus. The setting is available for Classification and Similarity solutions.
TF-IDF
T | F: TF-IDF is available for Clustering and Similarity solutions.
False - It is available for Classification and Similarity solutions.
What is the minimum number of required records for Classification solution training?
10,000
T | F: Clustering solutions can reuse an existing word corpus.
True
T | F: Clustering solutions groups data based on similarity.
True
T | F: Clustering solutions generate classes.
False
T | F: Clustering solutions does not support multiple languages.
False
What is the purpose of a word corpus?
The word corpus is the vocabulary the system uses to compare textual similarity between records.
What is the maximum classes a classification solution can support?
There is no limitation on the amount of classes a classification solution can support.
To predict field values, which type of solution should be created?
Classification
For classification solutions, how many possible outputs can a single solution include?
A solution can only have one output.
How do you preserve solutions during a system clone?
There is a system property that can be set to true to preserve your solutions.
