Automated Movie Performance Prediction

Abstract

In response to receiving the identifier of the movie, a prediction of the movie's performance is generated. Characteristics of the movie are identified, and a set of similar movies are identified based on the characteristics. Performance statistics are calculated for the set of similar movies indicating the average economic performance of those movies, and a prediction of the movie's performance is generated based on the performance statistics. The prediction is provided for presentation to a user.

Description

[Technical Field]

[0001] The described subject matter relates generally to computer-generated predictions, and more particularly to predicting movie performance. [Background technology]

[0002] This application claims the benefit of priority to U.S. Patent Application No. 16 / 883,316, entitled "Automatic Movie Performance Predictor," filed May 26, 2020, which is incorporated by reference.

[0003] The film industry is big business. Ticket sales at cinemas generated over $36 billion in revenue in 2016. Additionally, worldwide television and other post-theatrical video monetization will generate approximately $300 billion in future revenue. However, production costs are also high, with the average budget for producing a film in that same period well exceeding $100 million. While successful films have the potential to generate significant profits, unsuccessful films can result in equally significant losses. [Brief explanation of the drawings]

[0004] [Figure 1] FIG. 1 is a block diagram illustrating a networked computing environment in which movie performance predictions may be provided, according to one embodiment. [Figure 2] FIG. 1 is a block diagram illustrating a performance prediction system, according to one embodiment. [Figure 3] 3 is a block diagram illustrating a movie database of the performance prediction system shown in FIG. 2 according to one embodiment. [Figure 4] 2 is a block diagram illustrating an exemplary computer suitable for use in the networked computing environment of FIG. 1, according to one embodiment. [Figure 5] 1 is a flowchart illustrating a method for predicting movie performance, according to one embodiment. DETAILED DESCRIPTION OF THE INVENTION

[0005] The drawings and the following description describe specific embodiments for purposes of example only. Those skilled in the art will readily recognize from the following description that alternative embodiments of the structure and methods may be employed without departing from the principles described. Reference will now be made to certain embodiments, examples of which are illustrated in the accompanying drawings. It should be noted that, wherever practicable, like or similar reference numerals will be used in the figures to indicate like or similar functionality.

[0006] Overview and Benefits The success or failure of a given film can have a profound impact on a wide range of people and others. To name a few, actors' careers may be advanced or hindered, film studios may expand or close, and franchise owners may consider or abandon sequels. The perceived likelihood of a given film's success or failure may even determine whether the film itself is completed or shelved. Traditionally, stakeholders have often relied on the subjective opinions of critics and other human experts to predict a film's future performance. Furthermore, existing methods rely on proprietary or difficult-to-access data when generating predictions of a film's box office performance. Various embodiments are described that provide more objective predictions of a film's performance based on publicly (or at least more readily) available information.

[0007] In one embodiment, a computer-based method for predicting movie performance includes receiving an identifier for a selected movie and identifying a set of features for the selected movie. The method also includes identifying a set of similar movies based on the set of features, each movie in the set of similar movies having a characteristic indicative of similarity to the selected movie. Performance statistics are calculated for the set of similar movies indicative of economic performance of the movies in the set of similar movies. The method further includes generating a prediction of the performance of the selected movie based on the performance statistics and providing the prediction for presentation to a user.

[0008] In various embodiments, the disclosed methods can generate predictions of movie performance that are, on average, at least as accurate as those of experienced movie critics. These predictions are also based on objective data, rather than subjective judgments. Furthermore, these predictions may be generated based on readily available information. System example FIG. 1 illustrates one embodiment of a networked computing environment 100 in which predictions of movie performances may be provided. In the embodiment shown in FIG. 1, the networked computing environment includes a performance prediction system 110, one or more third-party servers 120, and several client devices 140, all of which are connected via a network 170. For illustrative purposes, three client devices 140 are shown, but any number of client devices may be connected to the network 170. In other embodiments, the networked computing environment 100 includes different and / or additional elements. Furthermore, functionality may be distributed among the elements in a manner different from that described. For example, in one embodiment, the performance prediction system 110 does not use any data from the third-party server 120. It should be noted that in some embodiments, the performance prediction system 110 is a standalone system and may not be connected to a network at all.

[0009] The performance prediction system 110 generates predictions regarding the performance of a particular film. The predictions may be based on publicly available information about the film before its release and, in some cases, before the film is completed. Examples of such information include the budget, genre, target demographics, associated intellectual property (e.g., copyrighted characters, the novel the film is based on, etc.), cast, goals (winning awards, making money, selling toys or other merchandise, building a brand, etc.), and whether the film is considered award-winning (e.g., likely Oscar nomination). In some cases, publicly available information may be supplemented with proprietary studio information. Thus, studios may be able to generate more accurate predictions for their films than those of their competitors. However, the performance prediction system 110 may generate accurate predictions using only publicly available information. Typically, predictions are generated before release and relate to future performance. Nevertheless, in some cases, predictions may be generated after the fact. For example, a user may want to compare the predictions generated by the performance prediction system 110 with the film's actual performance. Various embodiments of performance prediction systems are described below with reference to FIGS. 2 and 3.

[0010] Third-party server 120 is a computer system to which the performance prediction system may connect (e.g., via network 170) to obtain data or services. Although the term third-party is used, in some cases, third-party server 120 may be controlled by the same one that operates performance prediction system 110. In one embodiment, third-party server 120 hosts one or more databases of information about movies. These databases may include information such as title, release date (or planned release date), genre, cast, and budget. Performance prediction system 110 may connect to such information from one or more third-party servers 120 as needed. For example, performance prediction system 110 may submit a query including the movie title to third-party server 120 and receive the release date, genre, cast, and budget in response. In some cases, performance prediction system 110 may obtain different information from different third parties. For example, cast information may be obtained from one source while the budget is obtained from another source. Alternatively, as previously mentioned, some or all of this information may be stored by performance prediction system 110.

[0011] Client device 140 is a computing device capable of receiving user input as well as transmitting and receiving data over network 170. Client device 140 can take various forms, such as a desktop computer, a laptop computer, a personal digital assistant (PDA), a mobile phone, a smartphone, and other suitable devices. In one embodiment, client device 140 provides an interface (e.g., a web page presented in a browser, an app, etc.) through which a user may interact with performance prediction system 110. The user identifies a movie, and client device 140 queries performance prediction system 110 for a prediction of the identified movie's performance. The prediction may be generated in response to the query, or the performance prediction system 110 may retrieve a prediction for the identified movie in response to the query. Alternatively, a hybrid method may be used in which performance prediction system 110 generates a prediction for a movie when initially requested and then stores the generated prediction in a database. Thus, when performance prediction system 110 receives a query, it may check whether an existing prediction for the movie exists, and if not, generate a new prediction. Regardless of how the prediction is generated, the performance prediction system 110 transmits it to the client device 140 for presentation to the user.

[0012] Network 170 provides a communication channel through which other elements of networked computing environment 100 communicate. Network 170 can include any combination of local-area and / or wide-area networks, using both wired and / or wireless communication systems. In one embodiment, network 170 uses standard communication technologies and / or protocols. For example, network 170 can include communication links using technologies such as Ethernet, 802.11, WiMAX, 3G, 4G, Code Division Multiple Access (CDMA), and Digital Subscriber Line (DSL). Examples of network protocols used to communicate over network 170 include Multiprotocol Label Switching (MPLS), Transmission Control Protocol / Internet Protocol (TCP / IP), Hypertext Transport Protocol (HTTP), Simple Mail Transfer Protocol (SMTP), and File Transfer Protocol (FTP). Data exchanged over network 170 can be represented using any suitable format, such as Hypertext Markup Language (HTML) or Extensible Markup Language (XML). In some embodiments, all or part of the communication links of network 170 may be encrypted using any suitable technology or technique.

[0013] 2 illustrates one embodiment of performance prediction system 110. In the embodiment illustrated in FIG. 2, performance prediction system 110 includes movie data 210, a movie selection module 220, a feature extraction module 230, a similar movie identification module 240, a statistics generation module 250, and a prediction module 260. In other embodiments, performance prediction system 110 includes different and / or additional elements. Furthermore, functionality may be distributed among the elements in a manner different from that described. For example, rather than extracting features from movie data 210, some or all of the features may be obtained from third-party server 120.

[0014] Movie data 210 is information about movies stored on one or more computer-readable media. While movie data 210 is shown as a single entity, it may be stored on multiple devices in multiple locations. For example, movie data 210 may be stored in a distributed database connected by performance prediction system 110 over network 170. One embodiment of movie data 210 is shown in FIG. 3.

[0015] 3, movie data 210 includes information about several movies, Movie A 310 and Movie B 320 through Movie N 330. For each movie, movie data 210 includes a movie ID 312, features 314, and performance metrics 316. The movie ID may be a title, a unique identification number assigned to the movie, or any other data that can be used to identify a particular movie. When movie titles are used and multiple movies with the same title exist (e.g., remakes), additional information such as the release year may be included as part of the movie ID 312 to distinguish between the movies.

[0016] Features 314 are information about a movie that can be used to classify it and determine its similarity to other movies. In one embodiment, features include information available before release, such as budget, genre, target demographics, associated intellectual property, cast, and goals. Other examples of features include release date (e.g., movies released close to holidays are often similar), the origin of the underlying story (e.g., movies based on foreign folklore versus domestically generated stories), and whether the movie is considered to have award potential. Some or all of these features may be obtained from third-party services such as "BOX OFFICE MOJO," "IMDB," or "THE-NUMBERS.COM." The set of features 314 may be represented as a vector, called a feature vector, with an element holding a value for each feature in the set. It should be noted that in some embodiments, the performance prediction system may obtain some or all of the features 314 as needed from the third-party server 120, rather than storing them as part of the movie data 210.

[0017] Performance metrics 316 provide information about how a movie performed. In one embodiment, performance metrics 316 include worldwide box office, domestic box office, domestic opening weekend box office, and worldwide opening weekend box office. In other embodiments, performance metrics 316 may include different or additional indicators of a movie's performance. For example, for a movie made available for streaming or television, performance metrics 316 may include total views and revenue from these platforms.

[0018] Referring again to FIG. 2 , the movie selection module 220 provides an interface through which a user can select a movie for which a prediction is desired. In one embodiment, the movie selection module 220 provides a website accessible via the client device 140. The website includes a form that a user of the client device 140 fills out to select a movie. The form may solicit movie titles in free-text fields, provide search functionality based on one or more parameters (title, keywords, actors, director, genre, studio, etc.), or provide a list of movies (e.g., in a drop-down list). Alternatively, the form may ask for additional information about the movie to be used to generate the prediction. For example, the form may prompt the user to enter the cast (e.g., leading actors), budget, genre, target demographics, etc. Each characteristic may be entered via a free-text field, a drop-down list, etc.

[0019] In another embodiment, the movie selection module 220 provides an application programming interface (API) that allows software running on a client device (e.g., an app) to allow a user to select a movie. The interface provided by the software may operate in substantially the same manner as the various options described above for the website. In a further embodiment, the movie selection module 220 provides a similar interface in the performance prediction system 220. For example, if the performance prediction system 220 functions as a standalone system or is provided as a complete software package, the interface may be provided on the same device (e.g., client device 140) that generates the predictions.

[0020] The feature extraction module 230 identifies features of the selected movie that can be used to generate a prediction of the movie's performance. In one embodiment, the feature extraction module 230 queries a database (e.g., movie data 210) using the title or movie ID 312 of the selected movie for corresponding features 314. The feature extraction module 230 may alternatively or additionally collect information from one or more third-party servers 120 and / or the information may be provided by a user as part of the movie selection process. Regardless of the source or sources of information, the feature extraction module 230 extracts the desired features. The extracted features may be used to define a feature vector for the selected movie.

[0021] The similar movie identification module 240 compares the features of the selected movie with the features 314 of other movies (e.g., stored in the movie data 210) to identify a set of similar movies. Generally, two movies are considered similar if their corresponding features 314 are similar. For example, two movies featuring up-and-coming actors (e.g., Hailee Steinfeld, Tye Sheridan, Letitia Wright, Millie Bobby Brown, etc.) are generally more similar than a movie featuring one of those actors with an older movie star (e.g., Bruce Willis, Liam Neeson, Harrison Ford, Denzel Washington, etc.), all else being equal. As another example, two movies released during the week of Thanksgiving are more likely to be similar than a movie released that week and a movie released in the summer.

[0022] In various embodiments, the similar movie identification module 240 determines the distance between the selected movie and each movie in the movie data 210. In one embodiment, the distance is based on a set of similarity metrics. To calculate the similarity metric for a pair of movies, the similar movie identification module 240 compares the value of a particular feature of one movie with the value of the same feature of the other movie. The similarity metrics for a pair of movies may be combined (e.g., using a weighted combination function) to generate a single distance score that indicates the similarity between the movies. Thus, movies with similar values for many features are generally considered more similar than movies that share only a few similar features.

[0023] As a specific example of a similarity metric, to compare the lead actors in two films, the similar film identification module 240 can compare demographic information such as age, gender, and race with career statistics such as the number of previous films, the number of years since the first film, and the number of films in the last year to generate a difference score. For non-numeric data, the difference score may be based on a predetermined mapping between categorical values (e.g., a French actor may be considered more similar to a Belgian actor than a Chinese actor). The contribution of each factor considered may be weighted based on its significance. Thus, actors with similar demographic backgrounds and similarities in their careers will generally have a high similarity score (corresponding to a low distance), while actors from different demographic backgrounds with different career trajectories will generally have a lower similarity score (corresponding to a higher distance).

[0024] In another embodiment, the similar movie identification module 240 calculates the distance between the feature vector of the selected movie and the feature vector of each movie in the movie data 210. For non-numeric features, the distance between two values may be determined using a lookup table of predetermined distances. For example, for genre, the distance between science fiction and fantasy may be set to a first value (e.g., 1), while the distance between science fiction and documentary may be set to a second, higher value (e.g., 10).

[0025] In a further embodiment, the set of similar movies is selected using a machine learning model, such as a neural network. The machine learning model is trained before runtime by providing training data of human labels containing examples of similar and dissimilar movies. The labels may be binary (similar or dissimilar) or may indicate a similarity score (e.g., out of 5). The machine learning model is applied to the training data and updated (e.g., via backpropagation) until the output provided by the model matches the human-generated labels within a threshold tolerance. Once trained, the model uses feature vectors of the selected movie and the candidate movie as inputs. The model outputs a distance indicating how similar the selected movie and the candidate movie are.

[0026] Regardless of the particular method used, the similar movie identification module 240 selects the set of similar movies based on their distance from the selected movie. In one embodiment, a given movie is included in the set of similar movies if its distance is less than a similarity threshold. Alternatively, the set of movies may be of a fixed size, N, with the N movies with the lowest distance included in the set.

[0027] In some embodiments, rather than calculating the distance of each movie in movie data 210, similar movie identification module 240 performs a first filter based on one or more features. For example, only movies within the same (or closely related) genre may be considered eligible for inclusion in the set of movies. Similar movie identification module 240 then calculates a distance score for each movie that passes the filter. This may reduce processing power requirements when a large number of movies are included in movie data 260. Alternatively, the set of similar movies may be identified entirely through filtering by applying one or more filters.

[0028] Statistics generation module 250 calculates performance statistics based on the set of similar films. In one embodiment, the performance statistics include a budget-to-box office metric, a domestic percentage metric, and an average first weekend multiplier. In other embodiments, the performance statistics may include different and / or additional metrics. For example, metrics for a film may include the number of theaters that showed the film overall, overall revenue per theater, the number of theaters that showed the film during its opening weekend, revenue per theater during its opening weekend, the length of time the film was shown in theaters, the dates the film was shown in theaters, etc.

[0029] The budget to box office metric is a measure of the average ratio between a film's worldwide box office revenue and the budget of a set of similar films. The budget to box office metric may be calculated by dividing the worldwide box office revenue (e.g., as shown in performance metrics 316) by the budget of each film in the set (e.g., as shown in features 314), summing the resulting ratios, and dividing by the number of films in the set.

[0030] The domestic percentage metric is a measure of the average percentage of the worldwide box office revenue of the films in the set that consists of domestic (e.g., U.S.) box office revenue. In some embodiments, the performance prediction system 110 may provide a user interface that allows a user to select the countries that will be considered domestic for purposes of generating predictions. The domestic percentage metric may be calculated by dividing the domestic box office revenue by the worldwide box office revenue of each film in the set, summing the resulting percentages, and dividing by the number of films in the set.

[0031] The average first weekend multiplier is a measure of the films in a set's opening weekend and their overall performance. Historically, films were released on different days in different countries. Therefore, the average first weekend multiplier may be based on domestic performance. In such cases, the average first weekend multiplier may be calculated by dividing the total domestic box office receipts for each film in the set by the domestic box office receipts for the opening weekend, summing the resulting ratios, and dividing by the number of films in the set. However, as more and more films are released simultaneously worldwide (or at least in multiple countries), this ratio may additionally or alternatively be calculated from the total worldwide box office receipts and the first weekend worldwide box office receipts.

[0032] The prediction module 260 generates a prediction of the performance of the selected film based on the characteristics of the selected film and performance statistics for the set of similar films. In one embodiment, the prediction includes predictions of total box office, domestic box office, and opening weekend box office. The total box office may be predicted by multiplying the selected film's budget by the budget for a box office metric generated from the set of similar films. The domestic box office may be predicted by multiplying the predicted total box office by a domestic percentage metric generated from the set of similar films. The opening weekend box office may be predicted by dividing the predicted domestic (or worldwide) box office by a first weekend multiplier generated from the set of similar films.

[0033] In another embodiment, prediction module 260 may generate individual predictions for some or all foreign markets, rather than a single set of aggregated predicted foreign performance metrics. Predictions for a particular foreign market may take into account the prior performance of films of similar genres and franchises in that market, as well as the actors and other individuals involved. For example, a Hollywood film starring a foreign-born actor (even in a relatively minor role) may be predicted to perform better than other similar films in that country, especially if the actor is particularly popular in his or her home country. Conversely, the same film may perform worse than similar films in other countries that have poorer ties to the actor's country.

[0034] The prediction module 260 may generate different or additional outputs tailored to a particular use case. For example, in one embodiment, the prediction model 260 may predict demand for cross-sold merchandise for a movie. The prediction module 260 may automatically place orders for the merchandise, generate seller agreements, or create one or more marketing campaigns based on the predicted demand for the merchandise. As another example, the prediction model 260 may generate recommended cinema schedules for the movie (e.g., the number of screens on which to show the movie and the duration of the screening). As a further example, the prediction module 260 may determine when a movie should be released to optimize potential revenue, which may be performed manually or as part of an automated process for scheduling releases.

[0035] In yet another example, the prediction module 260 may output metrics for each role in a film (e.g., actor, director, writer, character, etc.) indicating each person's predicted impact on the film's revenue. In some cases (e.g., if an individual is expected to have a large impact on revenue), the metrics may be a dollar value or percentage increase (or decrease) in revenue. In other cases (e.g., if an individual's impact is predicted to be fairly small), the metrics may be relative to the role's average (e.g., the ratio of a particular individual's revision rate to the role's average revision rate). Thus, a studio can use the metrics to inform choices regarding who to hire for various roles in a project. In a further example, if the prediction model 260 has access to data for streaming films direct to video, it may output a recommendation to launch an on-demand or streaming version of the film alongside, before, or instead of the theatrical release. In other embodiments, the prediction may include different or additional estimates of the performance of the selected film.

[0036] Computing System Architecture 4 is a high-level block diagram illustrating an exemplary computer 400 suitable for use in networked computing environment 100 (e.g., as performance prediction system 110 or client device 140). Exemplary computer 400 includes at least one processor 402 coupled to a chipset 404. Chipset 404 includes a memory controller hub 420 and an input / output (I / O) controller hub 422. Memory 406 and a graphics adapter 412 are coupled to memory controller hub 420, and a display 418 is coupled to graphics adapter 412. Storage device 408, keyboard 410, pointing device 414, and network adapter 416 are coupled to I / O controller hub 422. Other embodiments of computer 400 have different architectures.

[0037] 4, storage device 408 is a non-transitory computer-readable storage medium such as a hard drive, compact disc read-only memory (CD-ROM), DVD, or solid-state memory device. Memory 406 holds instructions and data used by processor 402. Pointing device 414 is a mouse, trackball, touchscreen, or other type of pointing device and is used in combination with keyboard 410 (which may be an on-screen keyboard) to input data into computer system 400. Graphics adapter 412 displays images and other information on display device 418. Network adapter 416 couples computer system 400 to one or more computer networks (e.g., network 130).

[0038] 1-3 may vary depending on the embodiment and the processing power required by the embodiment. For example, movie data 260 may be stored in a distributed database system including multiple blade servers working together to provide the described functionality. Additionally, the computer may lack some of the components described above, such as keyboard 410, graphics adapter 412, and display 418.

[0039] Exemplary Methods FIG. 5 illustrates one embodiment of a method 500 for predicting movie performance. The steps of FIG. 5 are illustrated from the perspective of performance prediction system 110 performing method 500. However, some or all of the steps may be performed by others or components. Additionally, some embodiments may perform steps in parallel, in a different order, or perform different steps. For example, movie performance statistics may be pre-computed and stored (e.g., as part of movie data 210) with performance prediction system 110 accessing the movie performance statistics as needed.

[0040] 5, method 500 begins with performance prediction system 110 receiving 510 a movie selection. For the remainder of the description of method 500, it is assumed that the request is received from client device 140 for clarity and convenience. However, as previously mentioned, the request may also be provided via user input at performance prediction system 110.

[0041] The performance prediction system 110 identifies 520 features of the selected movie. For example, the features may include the budget, genre, target demographics, associated intellectual property, cast, movie goals, and release date. In one embodiment, the performance prediction system 110 checks whether the features of the selected movie are already available within the movie data 210. For example, if a user previously requested predictions for the same movie, the performance prediction system 110 may have stored the features at that time. If the features are not available in the movie data 210, the performance prediction system 110 may obtain them from a third-party server 120, solicit them from the user (e.g., by sending a request to the user's client device 140), or obtain them in any other suitable manner. Alternatively, the performance prediction system 110 may check one or more sources (e.g., the movie data 210 and a given third-party server 120) for the features and return an error to the client device 140 if the features are not available.

[0042] Based on the features of the selected movie, the performance prediction system 110 identifies 530 a set of similar movies. As mentioned above, the set may be of a fixed size (e.g., between 5 and 20 of the most similar movies, depending on system configuration or user-selected parameters), or it may include any movies that meet a predetermined metric, such as having a feature vector less than a threshold difference from the feature vector of the selected movie.

[0043] The performance prediction system 110 calculates 540 performance statistics for the set of similar films. For example, the performance statistics for each similar film may be the aforementioned budget-to-box office metric (total worldwide box office revenue divided by total budget), domestic percentage metric (domestic box office revenue divided by worldwide box office revenue), and average first weekend multiplier (domestic box office revenue divided by domestic opening weekend box office revenue). In one embodiment, the performance prediction system 110 retrieves data (e.g., budget and performance metrics 316) for each film in the set from the film data 210. The performance prediction system 110 calculates 540 performance statistics based on the retrieved data. For example, the performance prediction system 110 may calculate each of the aforementioned metrics for each film in the set using a set performance statistic that is the average value of each metric. Various types of averages may be used, including averages or weighted averages (e.g., weights assigned to each film in the set are based on that film's similarity to the selected film).

[0044] The performance prediction system 110 predicts 550 the performance of the selected film based on the performance statistics. In one embodiment, as described above, the performance prediction includes three components: total box office revenue, domestic box office revenue, and opening weekend box office revenue. Which market is considered the domestic market may be fixed or selected by the user. The prediction may be stored (e.g., within the movie data 210). In this way, if the performance prediction system 110 receives another request for a prediction of the performance of the selected film, the stored prediction may be used. Alternatively, a new prediction may be generated if certain metrics are met (e.g., each time a new request is received, if the stored prediction has been generated more than a specified time in the past, if more than a specified number of films have been added to the movie data 210, etc.). In such cases, a user interface may provide a user interface to allow a user (e.g., at the client device 140) to view multiple predictions, an average prediction, how the prediction has evolved over time, etc.

[0045] Regardless of the details of how it is generated, the performance prediction system 110 provides 560 the prediction for presentation. In various embodiments, the performance prediction system 110 sends data describing the prediction (e.g., the values of each element of the prediction) to the client device 140 from which the request originated. The client device 140 presents the prediction to the user. Alternatively, the prediction is sent to the user via an alternative communication channel relevant to the user. For example, the prediction may be sent to the user via email, instant message, text message, etc. In one embodiment, the user may select a delivery method for the prediction when requesting the prediction.

[0046] Additional Considerations Some portions of the foregoing describe embodiments in terms of algorithmic processes or operations. These algorithmic descriptions and representations are commonly used by those skilled in the data processing arts to effectively convey the substance of their work to others skilled in the art. While these operations are described functionally, computationally, or logically, they will be understood to be implemented by computer programs including instructions for execution by a processor or equivalent electrical circuits, microcode, or the like. Further, and without loss of generality, it is sometimes convenient to refer to arrangements of these functional operations as modules.

[0047] As used herein, a reference to "one embodiment" or "one embodiment" means that a particular element, feature, structure, or characteristic described in connection with that embodiment is included in at least one embodiment. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment.

[0048] As used herein, the terms "comprise," "comprising," "include," "including," "having," "having," or other variations thereof, are intended to cover a non-exclusive inclusion. For example, a process, method, provision, or apparatus comprising a list of elements is not necessarily limited to only those elements, but may include other elements not expressly listed in or inherent in such process, method, provision, or apparatus. Furthermore, unless expressly stated to the contrary, "or" refers to an inclusive or, not an exclusive or. For example, condition A or B is satisfied by any one of the following: A is true (or present) and B is false (or absent), A is false (or absent) and B is true (or present), and both A and B are true (or present).

[0049] Furthermore, the use of "a" or "an" is employed to describe elements and components of embodiments herein. This is done merely for convenience and to give a general sense of the disclosure. This description should be read to include one or at least one, and the singular also includes the plural unless otherwise clearly intended. When a value is described as "about" or "substantially" (or derivatives thereof), such value should be interpreted as being exactly + / - 10% unless otherwise clear from the context. From the example, "about 10" should be understood to mean "within the range of 9 to 11."

[0050] Upon reading this disclosure, those skilled in the art will recognize still additional alternative structural and functional designs for systems and processes for predicting movie performance. Thus, while specific embodiments and applications have been illustrated and described, it is to be understood that the described subject matter is not limited to the exact structure and components disclosed herein, and that various modifications, changes, and variations that will be apparent to those skilled in the art may be made in the arrangement, operation, and details of the disclosed methods and apparatus. The scope of protection is to be limited only by the following claims.

Claims

1. 1. A computer-implemented method for predicting movie performance, comprising: receiving an identifier for a selected movie for which performance statistics are not available; identifying a set of features of the selected movie; automatically identifying a set of similar movies by applying the set of features as input to a neural network that generates a distance score corresponding to each of a set of candidate movies, the distance score for a given candidate movie indicating a similarity between the given candidate movie and the selected movie, and candidate movies in the set of candidate movies are included in the set of similar movies in response to the distance score corresponding to the candidate movie satisfying a condition; calculating performance statistics for the set of similar movies, the performance statistics indicative of the economic performance of the movie within the set of similar movies; calculating an average economic performance for the films in the set of similar films based on the performance statistics, the average economic performance including an average budget-to-box office metric defined as the average of the budget-to-total box office ratios of each film in the set of similar films; generating a forecast of the economic performance of the selected films using the average economic performance, the forecast of the economic performance of the selected films including a predicted budget-to-box office metric defined as the ratio of the selected films' budget and predicted total box office revenue; providing the prediction for presentation; 10. A computer-implemented method comprising:

2. The computer-implemented method of claim 1, wherein the condition is that the distance score corresponding to the candidate movie is the lowest distance score of all distance scores calculated for the set of candidate movies.

3. The computer-implemented method of claim 1, wherein the condition is that the distance score corresponding to the candidate movie is less than a threshold.

4. 2. The computer-implemented method of claim 1, wherein the set of features includes at least one of budget, genre, target demographic, associated intellectual property, cast, goals, whether the film has the potential to win awards, or release date.

5. The steps of computing performance statistics include: retrieving performance data for each movie in the set of similar movies from a data store; calculating the average economic performance of the movies within the set of similar movies based on the performance data; The computer-implemented method of claim 1 , comprising:

6. The film performance data includes the budget and worldwide box office revenue, and the step of calculating the average economic performance includes: calculating, for each movie in the set of similar movies, the budget-to-box office metric by dividing the worldwide box office revenue by the budget; summing the calculated budget-to-box office metrics to generate a total; 6. The computer-implemented method of claim 5, further comprising: dividing the sum by the number of movies in the set of similar movies; and calculating the predicted budget-to-box office revenue metric for the selected movie by:

7. The film performance data includes worldwide box office revenue and domestic box office revenue, and the step of calculating the average economic performance includes: calculating a domestic percentage metric for each movie in the set of similar movies by dividing the domestic box office revenue by the worldwide box office revenue; summing the calculated national percentage metrics to generate a total; dividing the sum by the number of movies in the set of similar movies to obtain a predicted domestic percentage metric for the selected movie, wherein the predicted economic performance of the selected movie further comprises the predicted domestic percentage metric; The computer-implemented method of claim 5 further comprising:

8. the film performance data includes total domestic box office revenue and total domestic weekend box office revenue, and the step of calculating the average economic performance comprises: calculating a first weekend multiplier for each movie in the set of similar movies by dividing the total domestic box office revenue by the domestic opening weekend box office revenue; summing the calculated first weekend multipliers to generate a sum; dividing the sum by the number of movies in the set of similar movies to obtain a predicted first weekend multiplier for the selected movie, wherein the predicted economic performance of the selected movie further includes the predicted first weekend multiplier; The computer-implemented method of claim 5 further comprising:

9. 2. The computer-implemented method of claim 1, wherein the prediction includes at least one of: projected revenue for the film in each of a plurality of markets; cross-sold merchandise orders; a recommendation on the number of screens that cinemas should dedicate to the film; a recommendation on when to release the film; an indication of the contribution to total revenue of each of one or more roles associated with the production of the film; or a recommendation to launch the film on-demand or in parallel with a streaming service before or instead of a theatrical release.

10. 1. A non-transitory computer-readable storage medium comprising computer program code for predicting the economic performance of a movie that, when executed by a computing system, causes the computing system to perform operations including: receiving an identifier for a selected movie for which performance statistics are not available; identifying a set of features of the selected movie; automatically identifying a set of similar movies by applying the set of features as input to a neural network that generates a distance score corresponding to each of a set of candidate movies, wherein the distance score for a given candidate movie indicates a similarity between the given candidate movie and the selected movie, and candidate movies in the set of candidate movies are included in the set of similar movies in response to the distance score corresponding to the candidate movie satisfying a condition; calculating performance statistics for the set of similar movies, the performance statistics indicative of economic performance of the movie within the set of similar movies; calculating an average economic performance for films in the set of similar films based on the performance statistics, the average economic performance including an average budget-to-box office metric defined as the average of the budget-to-total box office ratios of each film in the set of similar films; generating a forecast of the economic performance of the selected film using the average economic performance, the forecast of the economic performance of the selected film including a predicted budget-to-box office metric defined as the ratio of the selected film's budget and predicted total box office revenue; providing said prediction for presentation; and A non-transitory computer-readable recording medium comprising:

11. A non-transitory computer-readable recording medium as described in claim 10, wherein the condition is that the distance score corresponding to the candidate movie is the lowest distance score of all distance scores calculated for the set of candidate movies.

12. 12. The non-transitory computer-readable storage medium of claim 11, wherein the set of features includes at least one of budget, genre, target demographic, associated intellectual property, cast, goals, whether the film has the potential to win awards, or release date.

13. Calculating performance statistics is retrieving performance data for each movie in the set of similar movies from a data store; calculating the average economic performance of the movies within the set of similar movies based on the performance data; 13. The non-transitory computer-readable storage medium of claim 12, comprising:

14. The film performance data includes the budget and worldwide box office revenue, and calculating the average economic performance includes: calculating, for each movie in the set of similar movies, the worldwide box office revenue by the budget to calculate the budget-to-box office revenue metric; summing the calculated budget-to-box office metrics to generate a total; and and dividing the sum by the number of movies in the set of similar movies.

14. The non-transitory computer-readable storage medium of claim 13, comprising:

15. The film's performance data includes worldwide box office revenue and domestic box office revenue, and calculating the average economic performance includes: calculating a domestic percentage metric for each movie in the set of similar movies by dividing the domestic box office revenue by the worldwide box office revenue; summing the calculated national percentage metrics to generate a total; and dividing the sum by the number of movies in the set of similar movies to obtain a predicted domestic percentage metric for the selected movie, wherein the predicted economic performance of the selected movie further comprises the predicted domestic percentage metric; The non-transitory computer-readable medium of claim 13, further comprising:

16. The film's performance data includes total domestic box office receipts and total domestic weekend box office receipts, and calculating the average economic performance includes: calculating a first weekend multiplier for each film in the set of similar films by dividing the total domestic box office revenue by the domestic opening weekend box office revenue; summing the calculated first weekend multipliers to generate a sum; dividing the sum by the number of movies in the set of similar movies to obtain a predicted first weekend multiplier for the selected movie, wherein the predicted economic performance of the selected movie further includes the predicted first weekend multiplier; The non-transitory computer-readable medium of claim 13, further comprising:

17. 1. A computer system for predicting movie performance, comprising: A non-transitory computer-readable recording medium containing executable computer program code, the computer program code comprising: receiving an identifier for a selected movie for which performance statistics are not available; identifying a set of features of the selected movie; automatically identifying a set of similar movies by applying the set of features as input to a neural network that generates a distance score corresponding to each of a set of candidate movies, wherein the distance score for a given candidate movie indicates a similarity between the given candidate movie and the selected movie, and candidate movies in the set of candidate movies are included in the set of similar movies in response to the distance score corresponding to the candidate movie satisfying a condition; calculating performance statistics for the set of similar movies, the performance statistics indicative of economic performance of the movie within the set of similar movies; calculating an average economic performance for films in the set of similar films based on the performance statistics, the average economic performance including an average budget-to-box office metric defined as the average of the budget-to-total box office ratios of each film in the set of similar films; generating a forecast of the economic performance of the selected film using the average economic performance, the forecast of the economic performance of the selected film including a predicted budget-to-box office metric defined as the ratio of the selected film's budget and predicted total box office revenue; a non-transitory computer-readable storage medium containing instructions for providing the prediction for presentation; a processor for executing said computer program code; A computer system comprising:

Images