Generating a recommended target audience based on determining a predicted attendance utilizing a machine learning approach

#Advancements in software and hardware platforms have led to a variety of improvements in systems that provide targeted digital communications to client devices. For example, digital communication distribution systems are now able to provide digital communications to many different client devices across computer networks based on various goals to move prospects through an interaction funnel including various stages such as awareness, interest, consideration, intent, evaluation, and purchase. Indeed, some systems can distribute digital communications for events of various types—in-person events, executive roundtables, trade shows, and webinars—to influence prospects through the interaction funnel toward a purchase.

#Despite these advances however, conventional digital communication distribution systems continue to suffer from a number of disadvantages, particularly in their accuracy, efficiency, and flexibility. Indeed, conventional systems inaccurately distribute digital communications to client devices of prospects by utilizing broad-based targeting techniques. For example, conventional systems often rely on targeting large numbers of prospects based on generic industry standard thresholds. These systems thus distribute digital communications to invite prospects to events in a broad sweep, with the general understanding that industry standards indicate that, on average, only a certain percentage of prospects who receive the digital communications will attend the event. In addition, many conventional systems are inaccurate in that the systems often fail to meet a particular event-related goal (e.g., a number of attendees or registrants) due at least in part to the non-specific nature of broad-based targeting, which in turn, results in higher churn.

#As another disadvantage, conventional digital content distribution systems are inefficient. In particular, many conventional systems inefficiently utilize computer resources such as computing time and computing power. To elaborate, due at least in part to the inaccuracy of these conventional systems, these systems often waste computer resources in generating and distributing digital communications to client devices of prospects that have a low probability of attending an event. Indeed, many conventional systems generate large amounts of spam communications that waste computing power and computing time by generating and distributing an ordinate number of digital communications with little or no chance of producing a result.

#This disclosure will describe one or more embodiments of the invention with additional specificity and detail by referencing the accompanying figures. The following paragraphs briefly describe those figures, in which:

#FIG. 1 illustrates an example environment for implementing an attendance optimization system in accordance with one or more embodiments;

#FIG. 2 illustrates an example overview of generating a recommended target audience in accordance with one or more embodiments;

#One or more embodiments described herein provide benefits and solve one or more of the foregoing or other problems in the art with an attendance optimization system that generates a recommended target audience for an event based on determining a predicted attendance utilizing an approximate nearest neighbor algorithm and an attendance prediction model (e.g., a deep neural network). While this disclosure refers to an attendance optimization system, the described system can be applied, not only for improving attendance, but also (or alternatively) for improving registration or other event-related key performance indices (KPIs). In some embodiments, the attendance optimization system can sequence one or more machine learning operations such as finding a set of individuals within a database that are similar to invitees for the event. The attendance optimization system can further generate predictions of likelihoods or probabilities that invitees will attend (or register for) the event. Based on the probability predictions of invitees attending the event, the attendance optimization system can further rank individuals identified as similar to the invitees according to the respective probabilities of corresponding invitees. In some embodiments, the attendance optimization system generates a predicted attendance (or registration) for the event based on the probabilities associated with the invitees and the corresponding similar individuals. Based on the predicted attendance being below a target attendance, the attendance optimization system can further generate a recommended target audience to achieve the target attendance. Thus, the attendance optimization system can use deep learning to determine both how many additional individuals to invite to an event and which additional individuals to invite to meet the target attendance.

#Based on analyzing a set of invitees that have been invited to attend an event (e.g., a roundtable, a video conference, a trade show), the attendance optimization system can identify similar individuals to also invite to the event. In some embodiments, the attendance optimization system identifies the similar individuals that, when invited to the event, will achieve or improve one or more KPIs. For instance, the attendance optimization system can generate a recommended target audience that includes the original set of invitees in addition to individuals within a threshold similarity of one or more of the invitees. Indeed, the attendance optimization system can add one or more of the similar individuals based on probabilities of the similar individuals attending the event.

#As mentioned, the attendance optimization system can find individuals within a database that are similar to invitees for an event. Indeed, the attendance optimization system can analyze a database of user information to identify users or individuals that are within a threshold similarity in relation to a set of invitees for an event. In particular, the attendance optimization system can utilize an approximate nearest neighbor algorithm to compare features associated with invitees within a set of invitees for a particular event with features associated with individuals whose information is stored within the database. In some embodiments, the attendance optimization system receives an indication (e.g., from an administrator device) to prioritize a feature, whereupon the attendance optimization system weights the feature more heavily in applying the approximate nearest neighbor algorithm.