How We Estimate Event Impact

Dr. Ali Gazala
Data Scientist

Demand Intelligence and Ranking

A continuous challenge for businesses around the world is to understand event impacts on business activities. Events are heterogeneous in nature and quantifying their influence in terms of ‘how’ and ‘why’ are crucial to business success. That’s where our demand intelligence platform came into play. It provides global event visibility that helps to generate actionable insights into what events might have impacted your business in the past, and what events are likely to impact your business in future. Therefore, you can ensure your customer service, marketing and pricing as well as logistics resources are allocated accordingly.

PredictHQ’s ranking algorithm is an essential part of our demand intelligence platform. It provides a systematic methodology for ranking events of various categories based on an underlying model to reflect the event impact and is a key component to our data enrichment process. Our ranking technology is a numerical value with a five-level hierarchical schema (minor=0-20, moderate=21-40, important=41-60, significant=61-80, and major=81-100). The ranking schema provides our customers with a simple quantitative value to compare events based on their estimated impact. For instance, a big concert for a pop star is likely to have a higher rank of 90 “major” than a local soccer match with a rank of only 45 “important”. Similarly, a “significant” natural disaster event in Europe should have the same damage impact as a “significant” heat wave in Australia. Thus, a substantial part of our research and development effort goes towards ensuring a standardized impact level across all event categories.

We've created three algorithms, each with a different focus:

  1. PHQ Rank™ - our core rank based on attendance

  2. Local Rank™ - impact relative to location

  3. Aviation Rank™ - impact relative to air travel

Our Data Approach – Hard Data vs Soft Data

Understanding the relationship between event size and its business impact is very important to our clients. To help achieve this goal, we collect and analyze hundreds of hard and soft data feeds to derive meaningful insights about events and their estimated impact. Hard data refers to quantitative values acquired through a systematic methodology; whereas soft data is based on qualitative and sentiment observations. For instance, historical attendance and venue sizes for scheduled events can be described as an objective hard data; while social media popularity and rating scores for music performers are examples of soft data. Estimating event impact requires a clever and deliberate combination of both hard and soft data sources. Therefore, PredictHQ’s ranking algorithm is optimized to emphasize the use of hard data over soft data to ensure our ranking can be used as an objective measure of impact.

Ranking Scale

PredictHQ’s API covers a wide range of different event types and sizes, like small meetups, mid-size music concerts, and large earthquakes with significant damage. Therefore, our ranking algorithm adapts a logarithmic scale that is designed to accommodate events of various impact sizes. The mathematical features of a logarithmic scale can insure our ranking is based on orders of magnitude to emphasize the relatively higher impact of large size events. For example, a small meetup event can be important to a local food delivery service, a national rugby match can impact the hospitality industry, and a large concert can strain the resources of public transportation or even air travel. End users can retrieve the relevant event sizes by adjusting the “rank_level” parameter in the PredictHQ API. Many of our customers use this feature to easily integrate event occurrences into further analysis such as demand forecasting and dynamic pricing.

We work hard to ensure our demand intelligence platform is continuously evolving in response to the demands and needs of our customers. We are currently focusing our R&D effort on developing a new ranking index that is more representative of the local event impact. The new index takes into account the population size to adjust the estimated event impact according to the local geographic area. For example, the annual World of WearableArt festival is a more significant event for a city the size of Wellington, New Zealand, as compared to a similar size event in New York city. Therefore, having a localized impact index will help our customers to refine their analysis at a more granular level.

PredictHQ’s mission is to be the single source of truth for events, providing global event visibility combined with reliable impact estimation through our demand intelligence platform. Having all the event data and intelligence you require wrapped up in one platform means PredictHQ does the heavy lifting, so you can focus on your core capabilities instead.