Data Types, Error Types, Analytics Types and Robocalls

This article provides an overview of important data, analytics types, and different error types. While these concepts apply to virtually any form of statistical analysis, the following focuses on the impact of analytics engines that seek to simultaneously provide consumer protection from unwanted robocalls while minimizing harm to legitimate businesses that seek to their clients and/or potential customers.

For example, the Do Not Originate (DNO) database is a repository of telephone numbers that may not be associated with call origination. This is an example of deterministic data as DNO data is based on telephone numbers of a certain type such as malformed numbers, non-NANPA numbers, non-allocated numbers, non-assigned numbers, or numbers that receive calls only.

DNO Telephone Numbers are an example of Deterministic Data

Therefore, use of DNO data is an example of how analytics engines may deterministically identify unwanted robocalls with an extremely low false positive rate. DNO data may be used with either content-based or event-based analytics.

Important Statistics Concepts for Robocall Analysis: False Positives and False Negatives

False Positives: Incorrect Identification of a Condition

In statistics, a false positive occurs when a test incorrectly indicates the presence of a condition or attribute that is not actually present. It's a type I error, where the test results in a positive outcome when it should have been negative.

In the context of robocalls, a false positive measurement from an analytics perspective would mean that the system incorrectly identifies a legitimate call as a robocall.

This could lead to the blocking or flagging of valid calls, causing inconvenience or missed important communications for users. Minimizing false positives is crucial in developing effective robocall detection systems to maintain the accuracy of call blocking or filtering mechanisms.

False Negatives: Wrongful Indication of Condition Not Present

A false negative in statistics occurs when a test incorrectly indicates the absence of a condition or attribute that is actually present. It's a type II error, where the test results in a negative outcome when it should have been positive.

In the context of robocalls, a false negative from an analytics perspective would mean that the system fails to detect a robocall, incorrectly allowing it to go through as if it were a legitimate call.

This situation is problematic as it can lead to users receiving unwanted and potentially malicious calls, undermining the effectiveness of the robocall detection system. Minimizing false negatives is essential to enhance the accuracy of identifying and blocking undesired robocalls.

False Positives and Missed Business Calls

False positives may cause consumers to miss important business calls such as:

• Job Interviews: Calls from potential employers regarding job opportunities.
• Client or Customer Calls: In the case of businesses, calls from clients or customers can be crucial for ongoing relationships and transactions.
• Medical Appointments: Calls from healthcare providers confirming or rescheduling appointments.
• Financial Transactions: Calls related to important financial matters, like verification calls from banks or credit card companies.
• Emergency Notifications: Calls from emergency services or official authorities in critical situations.
• Service Outages or Repairs: Calls from utility companies or service providers regarding disruptions or scheduled maintenance.

It's essential for call filtering or blocking systems to minimize false positives to ensure consumers don't miss these vital communications.

Content-based Analytics vs. Event-based Analytics

Content-based Analytics

Content-based analytics refers to the process of analyzing and extracting insights from the actual content of data. This approach involves examining the characteristics, patterns, and features within the content itself, rather than relying solely on metadata or external information.

In various fields, content-based analytics can be applied:

• Text Analytics: Analyzing textual content to extract information such as sentiment, key topics, or entities.
• Image Analysis: Extracting insights from visual content, which could include recognizing objects, patterns, or anomalies.
• Audio Analysis: Analyzing audio content for features like speech recognition, emotion detection, or identifying specific sounds.
• Video Analytics: Examining video content to derive information such as object recognition, activity detection, or scene analysis.

Content-based analytics is particularly valuable in gaining a deeper understanding of data, enabling more nuanced and context-aware insights across different types of content.

Event-based Analytics

Event-based analytics involves analyzing and deriving insights from specific occurrences or events within a system or dataset. Instead of continuously monitoring all data, this approach focuses on capturing and analyzing events that are significant or relevant to a particular context. The events may represent occurrences, transactions, changes in state, or other noteworthy activities.

Key aspects of event-based analytics include:

• Real-Time Analysis: Often, event-based analytics is associated with real-time processing, enabling immediate insights and actions based on recent events.
• Pattern Recognition: Identifying patterns and trends within a stream of events to make predictions or detect anomalies.
• Triggered Actions: Responding to specific events by triggering predefined actions, such as alerts, notifications, or automated processes.
• Contextual Understanding: Analyzing events within the broader context to gain a more comprehensive understanding of the underlying processes.

Event-based analytics finds applications in various domains, including finance, cybersecurity, Internet of Things (IoT), and business intelligence, where timely and context-aware insights are crucial.

Analytics Engines use both Deterministic and Stochastic Data

Analytics Engine use of Deterministic Data

Analytics engines use deterministic data to perform analysis based on known and certain information. Deterministic data consists of explicit values or facts without uncertainty. Here's how analytics engines leverage deterministic data:

• Precise Calculations: Deterministic data allows for precise calculations and predictions because the values are fixed and known. This is particularly important for applications where accuracy is crucial, such as financial forecasting or scientific research.
• Rule-Based Decision-Making: Analytics engines use deterministic rules and logic to make decisions or classifications. If-then rules are applied based on predefined conditions, allowing for deterministic outcomes.
• Querying and Filtering: In databases or analytics tools, deterministic data can be easily queried and filtered. This enables users to retrieve specific information or perform analyses with confidence in the accuracy of the results.
• Consistent Results: With deterministic data, analytics engines produce consistent and reproducible results. Given the same inputs and conditions, the engine will always yield the same output, providing reliability in analyses.
• Model Validation: Deterministic data is often used for validating and calibrating analytical models. It helps assess the accuracy of models and ensures they align with the known information.

Overall, deterministic data plays a crucial role in building reliable and accurate analytical models and systems. It forms the foundation for making informed decisions based on explicit and certain information.

Analytics Engine use of Probabilistic Data

Analytics engines make determinations based on stochastic data by employing probabilistic methods and statistical techniques. Stochastic data involves uncertainty and randomness, and analytics engines adapt to this by using probability distributions and modeling techniques. Here's how it's done:

• Probability Models: Analytics engines build models that represent the probability distribution of the stochastic data. This involves estimating the likelihood of different outcomes based on historical patterns or sample data.
• Bayesian Inference: Bayesian methods are commonly used with stochastic data. These methods update the probability distribution as new data becomes available, allowing the analytics engine to continuously refine its determinations.
• Monte Carlo Simulation: This technique involves running simulations using random samples from the probability distribution. By repeating the simulation many times, the engine can estimate the likelihood of different outcomes and make determinations based on the aggregated results.
• Machine Learning Algorithms: Machine learning algorithms, especially those in the realm of supervised learning and probabilistic models, are used to analyze and make determinations based on stochastic data. These algorithms learn patterns and relationships from training data to make predictions or classifications.
• Risk Assessment: Analytics engines assess the risk associated with stochastic data by considering uncertainty in decision-making. This is particularly important in fields such as finance and insurance.

By incorporating probabilistic thinking and statistical methods, analytics engines can handle stochastic data effectively, providing valuable insights even in situations where outcomes are not deterministic.

Analytics Engines use Data to Protect Consumers from Unwanted Robocalls

Various analytics engines and technologies are employed to protect consumers from unwanted robocalls. These systems use a combination of rule-based algorithms, machine learning, and real-time analysis to identify and filter out potentially harmful or nuisance calls. Some key components include:

• Call Blocking Apps: Many mobile devices use call blocking apps that incorporate analytics engines to detect and block robocalls. These apps often maintain extensive databases of known spam numbers and use real-time analysis to identify suspicious calls.
• Machine Learning Models: Advanced analytics engines leverage machine learning models trained on large datasets to recognize patterns associated with robocalls. These models can adapt and improve over time as they encounter new types of spam calls.
• Behavioral Analytics: Some systems analyze the behavioral patterns of calls, looking for characteristics commonly associated with robocalls. This includes call frequency, call duration, and patterns in the call initiation process.
• Whitelisting and Blacklisting: Analytics engines maintain lists of trusted (whitelist) and known spam (blacklist) numbers. Calls are evaluated against these lists to determine whether they should be allowed or blocked.
• Caller ID Verification: Enhanced caller ID verification systems use analytics to validate the authenticity of incoming calls. This helps in detecting spoofed numbers and preventing illegitimate calls.
• Voiceprint Recognition: In more advanced systems, voiceprint recognition technology is employed to analyze the audio content of calls, identifying known patterns associated with robocalls.

Various technologies and approaches are often leveraged as comprehensive solutions aimed at minimizing false positives and negatives, ensuring that consumers are protected from unwanted robocalls without missing important calls.

About the Author: Gerry Christensen

In his current role, Gerry Christensen is responsible for regulatory compliance as an internal advisor to Caller ID Reputation® and its customers as well as externally in terms of policy-making, industry solutions and standards. In this capacity, Gerry relies on his knowledge of regulations regarding B2C communications engagement. This includes the Truth in Caller ID Act, the Telephone Consumer Protection Act of 1991, state "mini-TCPA" laws and statutes governing consumer contact, various Federal Communications Commission rules, and the Federal Trade Commission's Telemarketing Sales Rule (FTC TSR)

Christensen coined the term, "Bad Actor's Dilemma", which conveys the notion that unlawful callers often (1) don't self-identify and/or (2) commit brand impersonation (explicit or implied), when calling consumers. These rules are addressed explicitly in the FTC TSR (see 310.3 and 310.4) and implicitly in the Truth in Caller ID Act. Christensen has expertise in VoIP, messaging and other IP-based communications. Gerry is also an expert in solutions necessary to identify unwanted robocalls as well as enabling wanted business calls. This includes authentication, organizational identity, and use of various important data resources such as the DNO, DNC and RND.

Gerry is also an expert in technologies and solutions to facilitate accurate and consistent communications identity. This includes authentication and validation methods such as STIR/SHAKEN as well as various non-standard techniques. His expertise also includes non-network/telephone number methods such as cryptographically identifiable means of verifying organizational identity. In total, Christensen's knowledge and skills make him uniquely qualified as an industry expert in establishing a trust framework for supporting wanted business communications.