Patent Application
20220036486
The present disclosure relates to deriving a rating for properties, and more specifically to applying data processing and learning methodologies to analyze property data for deriving property attractiveness ratings for properties globally.
Real estate has provided a vital part of the human experience. Real estate data may be used effectively to produce faster and better outcomes for all parties in the real estate value chain. Real estate data may be connected at a significant scale in both types and geographies. Buildings or property may be the natural anchors for data in the real estate sector, upon which hundreds of attributes may be collected and analyzed. For example, one attribute, the net operating income of a commercial building, can inform financial desirability for companies looking to acquire real estate assets. The proximity of food or transit options may inform how individuals or companies view location attractiveness of properties in an area. Many core building attributes such as square feet, year built, number of stories, and tax are critical to underwriting studies. Data in real estate may be complicated by the vastness of the real estate sector itself. Data in real estate may be measured based on the volume of transactions, the geographically wide asset distribution, the myriad of sources recording and originating data, and a highly complex and layered value chain.
For professionals in the industry or sophisticated buyers or investors, it may be a normal activity to gather the overall rating or score of a single property, building or a group of properties that may be of interest to the individual. Such a rating may be derived from a multitude of factors and variables and may not be fixed for an extended period of time. While several variations of solutions may exist in the marketplace, there is seldom a verifiable method to determine a consistent rating other than conventional wisdom or experience of a single individual or corporate entity. Moreover, since there isn't a verifiable and deterministic method of arriving at such a rating, it is difficult to maintain the provenance of the rating and making such ratings susceptible to legal contention may cause an unnecessary investment and not make the projected return.
In accordance with some embodiments of the present disclosure, there is provided a computer-implemented method for analyzing property data to deriving property attractiveness ratings for properties globally. The method can include providing a Property Attractiveness Rating (PAR) application executed by one or more processors of a computing system. The method can include executing the PAR application to cause the computing system to perform operations of storing data associated with a plurality of properties database, and presenting a geographical map-based interactive user interface on a display of a user computing device for a user to search for at least one property on a map. The method can also include executing the PAR application to cause the computing system to perform operations of receiving a user input with one or more property attributes via the geographical map-based user interface, and searching the database to obtain and present one or more properties of interest on the map based on the user input. The method can further include executing the PAR application to cause the computing system to perform operations of receiving a user selection of a searched property of interest on the map via the geographical map-based user interface, generating a PAR for the selected property of interest, and presenting the generated PAR of the selected property of interest with respective property information on the map on the geographical map-based user interface.
Furthermore, in accordance with some embodiments of the present disclosure, there is provided a computing system comprising one or more processors, and one or more non-transitory computer-readable storage devices storing computer-executable instructions. The computing system may execute a Property Attractiveness Rating (PAR) system being stored as the computer-executable instructions. The one or more processors, when executing the instructions, causes the computing system to perform operations of storing data associated with a plurality of properties in a database, and presenting a geographical map-based interactive user interface on a display of a user computing device for a user to search for at least one property on a map. The one or more processors, when executing the instructions, causes the computing system to perform operations of receiving a user input with one or more property attributes via the geographical map-based user interface, and searching the database to obtain and present one or more properties of interest on the map based on the user input. The one or more processors, when executing the instructions, causes the computing system to perform operations of receiving a user selection of a searched property of interest on the map via the geographical map-based user interface, generating a PAR for the selected property of interest, and presenting the generated PAR of the selected property of interest with respective property information on the map on the geographical map-based user interface.
Furthermore, in accordance with some embodiments of the present disclosure, there is provided a computer program product comprising one or more computer-readable storage devices having a Property Attractiveness Rating (PAR) system encoded as computer-executable instructions. The computer program product, when executed by one or more processors of a computing system, causes the computing system to perform operations of storing data associated with a plurality of properties in a database and presenting a geographical map-based interactive user interface on a display of a user computing device for a user to search for at least one property on a map. The computer program product, when executed by one or more processors of a computing system, causes the computing system to perform operations of receiving a user input with one or more property attributes via the geographical map-based user interface and searching the database to obtain and present one or more properties of interest on the map based on the user input. Further, the computer program product, when executed by one or more processors of a computing system, causes the computing system to perform operations of receiving a user selection of a searched property of interest on the map via the geographical map-based user interface, generating a PAR for the selected property of interest; and presenting the generated PAR of the selected property of interest with respective property information on the map on the geographical map-based user interface.
The accompanying drawings, which are incorporated in and constitute part of this specification, and together with the description, illustrate and serve to explain the principles of various example embodiments. In addition, it should be understood that any figures which highlight the functionality and advantages are presented for example purposes only. The disclosed methodology and system are each sufficiently flexible and configurable such that they may be utilized in ways other than that shown.
The present description is made with reference to the accompanying drawings, in which various example embodiments are shown. However, many different example embodiments may be used, and thus the description should not be construed as limited to the example embodiments set forth herein. Rather, these example embodiments are provided so that this disclosure will be thorough and complete. Various modifications to the exemplary embodiments will be readily apparent to those skilled in the art, and the generic principles defined herein may be applied to other embodiments and applications without departing from the spirit and scope of the disclosure. Thus, this disclosure is not intended to be limited to the embodiments shown, but is to be accorded the widest scope consistent with the principles and features disclosed herein.
Embodiments of the present disclosure describe systems and methods of deriving ratings for buildings or properties. Embodiments described herein may facilitate the derivation and distribution of property attractiveness ratings for real estate buildings and properties across the world where public and private records may exist, by utilizing sophisticated processing systems and multiple analytics modules of a Property Attractiveness Rating (PAR) system.
According to the embodiments of the disclosure, an interactive application may be provided and delivered to an end user either as a mobile application 135 running on a computing device 130 or a web browser application 134, which allows the user via a user device to select a particular area or a singular point of interest on a geographical map-based interactive user interface. Further, the user may interact with the selected area or points of interest via the interface to review a property attractiveness rating which is delivered using a learning system with a set of modules. A learning system may be trained to learn based on several inputs and select the appropriate algorithm from a template of algorithms to derive and generate the property attractiveness rating.
In this context, a Property Attractiveness Rating (PAR) system may encapsulate a singular process to ingest several variables in real time and produce a single output or an outcome that embodies the rating of a building or a property. The PAR system may be established and deployed to provide real time solutions in response to user requests for property ratings.
Embodiments described herein address a practical computer-centric problem of providing a property attractiveness rating system for deriving and generating a property attractiveness rating for a building/property based on property data description. Further, embodiments described herein may provide technology-based improvements for collecting and transforming proprietary and third-party data through a multi-layered, complex data system and resolving non-value adding complexity within data into a far more usable property attractiveness rating (PAR) by utilizing a combination of property data analysis and machine learning techniques.
The embodiments of the present disclosure PAR system may be implemented as one or more computer programs or as application software executed on one or more computing devices that process property data of the business entity to provide property rating solutions. The improvements of the present disclosure as reflected in embodiments described herein may generate an automated property rating system to provide real time solutions for each respective entity.
As used herein, the term “Property Attractiveness Rating (PAR)” may include, but are not limited to, a PAR score, a property rating report, and any description for describing rating and comparable information associated with one or more properties and/or buildings.
As used herein, the term “model,” “learning model,” or “machine learning model” may include any type of a state-of-the-art model such as linear models and non-linear models.
Application server 102 may include an example property attractiveness rating (PAR) system/application 101, a proprietary data store 110 (e.g., database), and other program modules 200 which are implemented in the context of computer-executable instructions and executed by one or more processors 111 of application server 102. The PAR system 101 may be implemented in the context of computer-executable instructions and executed by one or more processors 111 for providing PAR computer-hosted services and/or providing a website with PAR services for a user to visit via a browser or a mobile application running on a computing device 130. Application server 102 may host an example property attractiveness rating system/application 101 which a user may access using a user computing device 130 via a browser application 134 or a mobile application 135 via network 150.
The PAR system 101 may be configured to analyze property data 114 and generate a property attractiveness rating. The PAR system 101 may include various program modules 200 that which may provide interactive mapping, visualization, searching and viewing mechanisms implemented by hardware, software, or combinations of hardware and software. Details related to program modules 200 of the PAR system 101 will be described below.
Data aggregation module 120 may aggregate data associated with real estate buildings or properties from different resources via network 150. For example, data may be sourced from existing internal master data stores, expert user inputs, appraiser inputs, 3rd party external platforms, crowd sourced inputs, and other available property stores. Data aggregation module 120 of the system 100A may be configured to retrieve and aggregate data through API calls via network 150 from various resources: 1) existing or internal property data stores; 2) external (third-party) real estate data stores and/or services; 3) data inputs by human professionals such as experts, appraisers or market researchers; and 4) crowd-sourced data inputs via the PAR system 101, and 5) other available property stores. Throughout this disclosure, example API calls are shown. It should be understood that any API calls shown are presented for example purposes only and are not intended to be limited to the APIs shown. The disclosed system is flexible and configurable such that other APIs may be utilized in ways other than that shown to request and retrieve additional data from the API. An example API call is illustrated below and may include a plurality of features or attributes associated with a building or a real estate property. For example, the plurality of features or attributes associated with the building or a real estate property may include property information, such as “address,” “city,” “zip code,” “latitude,” “longitude,” “population within 5 mile,” “transit score,” and “walk score,” etc.
Proprietary data store 110 may capture and store property data obtained by the data aggregation module 120 over the network 150. Proprietary data store 110 of the example system 100A may be a shared remote database, a cloud database, or an on-site central database. Proprietary data store 110 may be included in the application server 102, or coupled to, or in communication with, the application server 102 via network 150. Proprietary data store 110 may be an internal data store which may capture and store information relevant to display of property information as well as the property attractiveness rating generated and provided by the PAR system 101. Proprietary data store 110 may receive instructions or data from, and send data to, the PAR system 101. The property data associated with each building or real estate property may include a unique property identifier (ID), property images, a Property Attractiveness Rating (PAR), and a plurality of features or attributes, and any other property details or information. Proprietary data store 110 may be continuously updated with the property data which is aggregated, received, and generated by the PAR system 101.
Referring to
A user may access services of the PAR system 101 hosted by an application server 102 in a computing system 100A as described in
Network 150 may be either internal (intranet) or external (internet) or other public or private networks or combinations thereof. The end user can access the PAR system/application 101 on the corporate intranet or via an internet connection outside the company firewall. Network 150 may allow the visual content, search results, property details and property attractiveness ratings to be delivered to a user computing device 130 operated by an end user.
A Base Map Layer 210
A base map layer 210 in the PAR system 101 may be configured to present the geographical map-based interactive visual interface of the area or region around the search results in response to a user search request. In its default state, the base map layer 210 may determine the current user location and present a view within a preset radius around the user's location on a user interface of the user device 130 operated by the user. The user's location may be determined via the cellular connection, Wi-Fi positioning system (WPS) or GPS position of the user device 130. The base map layer 210 may bootstrap a map customized to the form factor of the user computing device 130 to provide a consistent visual experience to the user. The base map layer 210 may provide the initial scaffolding configuration for the map visualization and may contain only the schematic of the area of interest to the user.
A PAR Data Layer 220
The PAR data layer 220 may be configured to pull data from the proprietary data store 110 and prepare the data to be integrated into the base map layer 210. The PAR data layer 220 may contain the modules illustrated in
An Advanced Visualization Layer 230
The advanced visualization layer 230 may provide an interactive layer presented to an end user computing device 130. The advanced visualization layer 230 may provide an end user/client with a front-end interface such as a mobile application 135 or a browser application 134 to interact with the PAR system 101. The advanced visualization layer 230 may ingest inputs from the PAR data layer 220 and provide a visual of the underlying data as it is pinned on the base map layer 210.
The advanced visualization layer 230 may be configured to enable a computing device 130 operated by an end user/client to perform operations, including:
Property Details Module 310
The property details module 310 may present specific property information such as, but not limited to:
PAR Details Module 320
The PAR details module 320 may provide a dynamically refreshed or updated property attractiveness rating retrieved from the PAR data layer 220. The property attractiveness rating may be continuously updated as a result of changes being made into the PAR data layer 220 due to the updated data ingestion from multiple sources including third-party, expert and crowd-sourced inputs. Both property details module 310 and PAR details module 320 may contain logic and visualization components to present relevant information to the user device 130. The visualization components may adapt to either touch inputs, gesture inputs or computer mouse-based inputs.
Data Processing Module 410
After the property data is obtained and aggregated by a data aggregation module 120, and stored in the proprietary data store 110, some fundamental and data preprocessing may be performed. Data processing module 410 may fetch the data from the proprietary data store 110 and cleanse the data. Data processing module 410 may perform actions such as fetching data from the data store, cleansing data, removing duplicates or irrelevant/incomplete records, creating a training set and testing the data received. The actions may be continuously performed as a background task and may not interfere or coincide with a user's interactions with the advanced visualization layer 230. Data processing module 410 may interact with a learning system module 420 which may select the best fit rating algorithm from a template of algorithms.
Learning System Module 420
Learning system module 420 may interact with system evaluation module 430 which may calculate the rating and create a final algorithmic model to determine the rating based on the data gathered so far. The system evaluation module 430 may update the proprietary data store 110 with the calculated rating for future consumption by property details module 310 and the PAR details module 320. Details related to learning system module 420 of the PAR system 101 will be described below.
System Evaluation Module 430
System evaluation module 430 may be a dynamic component and may react to any expert or crowd sourced data inputs from the advanced visualization layer 230. Any new data may be rated against the final model to determine the property attractiveness rating. System evaluation module 430 may create a final model and update the PAR score based on the updated property data in the proprietary data store 110.
During the process 500, an interactive PAR application/system 101 may be provided and delivered to the user device 130 either as a mobile application 135 running on a mobile device or web browser application 134. The interactive PAR application/system 101 may allow the user to first select a particular area or singular point of interest in an intuitive manner. Further, by interacting with the selected area or points of interest, the user may review a property attractiveness rating which is delivered and/or generated using a set of modules. The set of modules of the learning system module 420 may learn based on several inputs and select the appropriate algorithm from a template of algorithms to deliver the rating. The model learning and selection methods may be deterministic and verifiable.
At step 502, the property data or information of a plurality of properties may be stored in a database (e.g., proprietary data store 110). The database 110 may ingest and store data aggregated from a data aggregation module 120 for various resources. The property data or information for each property may include a plurality of property attributes or features. Referring to
At step 504, the advanced visualization layer 230 of the PAR system 101 may present a geographical map-based interactive user interface on a display of a user device 130 for a user to search for at least one property on an electronic map.
At step 506, the PAR system 101 may receive a user input with one or more property attributes via the geographical map-based user interface which enables the user to navigate, zoom, enter and search property data. For example, the PAR system 101 may be executed to display the user interface components allowing the user to enter a property address or zip code to search for the property and corresponding searching the property attractiveness rating(s) of one or more properties in an area. As illustrated in a screenshot of 602 of
At step 508, responsively to the user input, the PAR system 101 may search the database 110 to obtain and present one or more properties of interest on the map. For example, in response to a user input such as a property address or a zip code, the PAR system 101 may present one or more properties of interest in the area associated with a property address or a zip code on the map on the geographical map-based interactive visual interface. In some embodiments, the PAR system 101 may receive new data associated with one or more properties which are not stored in the database 110 via a data ingestion process. The new data may be received through geographical map-based user interfaces associated with a plurality of user computing devices 130 operated by users. The PAR system 101 may dynamically update the data in the database 110 with the new data.
At step 510, the PAR system 101 may receive a user selection of a searched property of interest on the map via the geographical map-based user interface. The geographical map-based user interface may enable the user to filter the searched property data, navigate, zoom to the property area and select a searched property of interest displayed on the map via the geographical map-based user interface. For example, as illustrated in a screenshot of 604 of
At step 512, the PAR system 101 may generate a PAR based on the property data of the selected property of interest. The PAR system 101 may utilize data processing module 410, learning system module 420 and system evaluation module of PAR data layer 220 which are executed by a hardware processor to derive and generate a PAR for the selected property of interest. The derived and generated PAR may include a PAR score, a property rating report, or combinations thereof. The derived and generated PAR may include property details of one or more related properties and or buildings.
At step 514, the derived and generated PAR of the selected property of interest with respective property information may be presented on the map via the geographical map-based user interface. The PAR system 101 may utilize the property details module 310 and the PAR details module 320 to present the derived and generated PAR on the user interface. As illustrated in a screenshot of 608 of
In the process 500, data processing module 410, learning system module 420 and system evaluation module 430 of the PAR data layer 220 of the PAR system/application 101 may be replicate to derive the PAR score in response to user requests if necessary. The PAR system/application 101 may be executed by the one or more processors 111 to derive a PAR score through a sequence of processes. For example, a data ingestion process may be executed to receive the data associated with a plurality of properties into the database. A data processing process may be executed to process the data to create a training dataset and a test dataset. A learning process may be executed to train a plurality of models of a learning system in the PAR system with the training dataset and the test dataset. An evaluating process may generate a final model from the trained models. The evaluating process may be executed to apply the final model to the selected property of interest to generate the PAR of the selected property of interest.
The data processing module 410 of the PAR data layer 220 may perform actions to create property feature data based on the data stored in the database 110. Data processing module 410 of the PAR system may be executed by one or more processors 111 of application server 102 to implement a data processing process to perform actions such as fetching data from the data store, cleansing data, removing duplicates or irrelevant/incomplete records, creating a training set and testing the data received.
Fetch Data, Cleanse and, Split it into Training and Test Datasets
Fetching data, cleansing data, and splitting data into training and test sets will now be described with reference to
The data cleansing phase may involve looking for missing values and deciding whether to fill them in or leave them as is, converting all values into a numerical format, for instance, converting true/false into I/O or binning a range of values into categories. For example, data processing module 410 may take a count of missing values across all variables and rank them. And then, it might fill any missing values in a variable(s) with the most frequent type or its mean, etc. Thus, the data may be as complete as possible.
Data processing module 410 may encode all categorical variables and convert any remaining non-numerical variables into numerical factors. One way to do this is to generate dummy levels for each data observation. As an example, there are three Property Types: Commercial, Residential and Mixed. For each observation, data processing module 410 may generate a matrix of these three levels like:
A property can only be of one type. For example, the first one is Commercial, the second is Mixed and so on. A non-trivial matrix or a set of matrices may be used to represent the linear or non-linear combination of the variables. A matrix [m, n] is an array of m rows and n columns wherein a matrix of one row or one column is called a row vector and column vector, respectively. The matrix may help compute in an efficient and reliable way the estimates for each variable and then the combination of such to derive the rating of the property.
The data preprocessing may include marking any duplicate data with flags such that the duplicate data is not be processed any further, checking for the validity of the data to identify incorrect data types such as strings having unexpected numbers or invalid date formats such as dd-mm-yy. Further, the data processing module 410 may perform data cleansing such as removing duplicates and correcting invalid data in the data aggregation module 120.
The data processing module 410 may perform preprocessing operations on the cleansing data to construct data features. An example method of such a data transformation may be outlined below:
In some embodiments, data processing module 410 of the PAR data layer 220 may receive several variables from different systems as inputs and analyze them to select the essential variables. The essential variables may be used to define and construct the internal and external features of a building or a property. A variable may be a factor or an attribute about the property or a building that may affect the outcome of the PAR system.
The defined and constructed features may be used to call out corresponding variables. The outcome may be represented by a combination of a plurality of features that essentially facilitate understanding the relationship between the features and the outcome. The outcome may depend on the relationship among the features and may be called a dependent variable. The set of features that yield the dependent variable or the system outcome may be independent and called independent variables. Each of these independent variables may have an estimate associated with each of them. The sum of products of respective independent variables and respective estimates may reliably produce the outcome or the dependent variable.
Once all the variables (e.g., features) have been preprocessed, the data processing module 410 may fetch the data from the proprietary data store 110 to select features that are essential for training and building the learning system module 420 for determining the PAR score 708 shown in
The described example method and other similar methods may be applied to features to facilitate narrowing down and selection of the features that are essential to determine the PAR score 708.
Once the features or feature data are selected as a result of the operation of the data processing module 410, a series of models (e.g., machine learning models) may be tested to determine which models fit the selected feature data the best to help compute the PAR score. The models may be any type of state-of-the-art model, such as linear models or non-liner models. Data processing module 410 may interact with a learning system module 420 which may select the best fit rating algorithm from a template of algorithms.
In some embodiments, learning system modules 420 may include a host of methods and algorithms that help infer a function ƒ that maps input variables χ in such a way that the mapped function ƒ(χ) generates the data and generalizes to any new input variables. The described method may be called Regression.
An Outline of Regression
Regression models the relationship between a dependent variable or an outcome and one or more predictor variables (e.g., independent variables). A simple example in commercial real estate may be how access to public transport (independent variables) affects the asking price of a building (dependent variable or outcome). This linear relationship between the asking price (dependent variable Y) and the access to public transport (independent variable X) can be defined as:
Y=b
0
+b
1
X
Where b0 is known the intercept or the constant and b1 is the slope of X. In some embodiments, a Multiple linear regression (MLR) method may be used to model the linear relationship between one or more independent variables and a dependent variable. For example, a set of predictor variables or independent variables may be denoted as X1, X2, X3 and so on. In that case, the equation of the linear relationship transforms into:
Y=b
0
+b
1
X
1
+b
2
X
2
+b
3
X
3
This equation that defines the model tries to find the best line to predict the asking price as a function of the set of predictors. The data may not exactly fall on the line. Thus, the equation may include an error term:
Y=b
0
+b
1
X
1
+b
2
X
2
+b
3
X
3+ε
Because a line may be fitted onto the data, fitted values or estimates are obtained as opposed to the original known values. Thus, the target equation may become:
{circumflex over (Y)}={circumflex over (b)}0+{circumflex over (b)}1X1+{circumflex over (b)}2X2+{circumflex over (b)}2X2+{circumflex over (b)}3X3
The residuals may then be the difference between the fitted values and the known values:
εi=Yi−{circumflex over (Y)}i
The regression line which helps fit the data is then essentially the estimate that minimizes the sum of squared residuals (squaring helps solve for any negative values affecting the overall sum). In some embodiments, an Ordinary Least Squares regression method may be used to find the regression line of best fit for a set of data.
Diagnostics 704
After the feature data is modeled through a regression equation, feature data accuracy and robustness need to be checked by using different diagnostics methods before training a model. As shown in a block of diagnostics 704 in
Assumptions Checks
The PAR system 101 may check how normally distributed PAR scores are. As normal distribution is a prerequisite for regression, with mean zero and variance one, it is extremely important to normally distribute data sets that are not otherwise normally distributed. For example, the PAR system 101 may apply transformations such as Log Transform, Box Cox, etc, which may yield a normally-distributed data set.
Another way to check is to take n number of samples, then the t-statistic may follow Student's t-distribution with n−1 degrees of freedom.
Outlier Tests
Standardized residuals, or the number of standard errors away from the regression line, define the outliers and are described below. They are plainly the data that is far from the usual values or out of range. Usually, such data is dropped after the impact of such elimination on the model is assessed and neutralized.
Correlation and Variance Checks
The PAR system 101 may check for correlations of all variables with PAR score 708, and between the features as well. This helps eliminate variables that are already correlated with others, and thereby helps select the best variables that describe the data.
Variance in data may be another check that needs to be done. For instance, the PAR system 101 may check why a range of values have more errors than another range thereby making the model incomplete. If such large variance exists, then the data range is smoothed to minimize the impact of variance.
K-Fold Cross-Validation
The preprocessed feature data or datasets may be used to train and build a model of the learning system module 420. The feature data may be split into a training dataset and a testing dataset. There is a possibility that the testing dataset is easier or very difficult. An alternative is cross-validation, a way to evaluate how well the model may perform on data that it has not seen before. In K-fold Cross Validation, the dataset is split into K sections/folds where each fold is used as a testing set at some point. Assuming the dataset is split into K=5 folds, in the first iteration, the first fold or ⅕th of the data is used to test the model and the rest to train. In the second iteration, the second fold is used for testing and the rest for training. This process may be repeated until each fold of the five folds have been used as the testing set.
The learning system module 420 may represent a learning system in the PAR system and include a plurality of models. A learning process may be executed by one or more processors 111 of application server 102 to train the plurality of the models. The plurality of the models may be trained in an optimal way such that it produces an outcome from the combination of independent variables with near-consistent results. The training process may involve taking a sizable sample of the features that may be tuned by usually, but not necessarily, weights or parameters so that the model begins to yield near-consistent outcomes. In some embodiments, the learning process may be executed to apply multiple regression models to determine a combination of attributes of a property which are well-suited to derive a PAR score for a property. The learning process may be executed to select a Multiple Linear Regression (MLR) model that involves a linear combination of variables that are linearly independent and normally distributed. In some embodiments, the learning process may be executed to apply a Polynomial Regression model that involves a non-linear combination of variables that are linearly independent and normally distributed. In some embodiments, the learning process may be executed to apply a Weighted Regression model that involves a linear or non-linear combination of variables that are linearly independent and normally distributed with weights assigned to each variable as determined to be necessary.
Once the model is trained enough to produce outcomes that are repeatable, the model is tested against that sample space that was left out from the training phase. The test data may be chosen such that its features embodying a property already have a rating, which can be tested against the outcomes when the model is applied on the test data. In some embodiments, a hard-code processing may need at least one thousand properties, across geographies, property types, and price buckets, with a PAR score to be used as the baseline.
Diagnostics 706
With the checks described above, a model performance may be evaluated further to assess its accuracy. As shown in a block of diagnostics 706 in
Root Mean Square Error (RMSE) represents the sample standard deviation of the residuals calculated as:
Mean Absolute Error (MAE) is the average of the absolute difference of the residuals calculated as:
RMSE differs from MAE in the sense that it penalizes any large variance between the predicted and known values or if the residuals are larger.
R-Squared may be used as a statistical measure in a regression model that determines the proportion of variance in the dependent variable that may be explained by the independent variable. R-squared may show how well the data fits the regression model.
Prediction Intervals may be used to assess the model and determine how much confidence can be placed in a predicted value by determining the uncertainty around it. More precisely, after a sufficient number of values is recorded, a prediction interval is an estimate of an interval in which a future value will fall, with a certain confidence level. For instance, a prediction interval of 95% for a value or a range of values would mean that as more values are being recorded, this new value or the range of values may occur in 95% of such cases.
An evaluating process may be executed by one or more processors 111 of application server 102 to evaluate and select the appropriate and/or best algorithm from a plurality of trained algorithms/models based on the measures of diagnostics 706. With one or more trained models, the trained learning system module 420 may interact with the system evaluation module 430 which creates a final model to determine a PAR score 708 for a property in response to user inquires. The final model may be generated based on the updated data in the database 110. The final model may be applied to the selected property of interest to generate the PAR of the selected property of interest. The evaluating process may be executed to update the generated PAR of the selected property of interest in the database 110. In some embodiments, the property attractiveness rating may manifest as a point on a scale, or as a classification.
The PAR (score) 708 may be delivered and presented to a user on a geographical map-based interactive user interface over the network 150 via the advanced visualization layer 230 of the PAR system 101. A user may be able to deconstruct the property attractiveness rating to a certain degree and be able to understand the constituent driving factors behind a given building rating. A user may potentially be able to toggle and/or adjust assumptions on a subset of variables that are consequential to the rating calculation and observe the resulting rating differential. That is, by leveraging systems and models, the property attractiveness rating may provide a broad set of industry players a viewpoint on the relative attractiveness of a building vs different levels of geographical groupings (e.g., the immediate surroundings, the district, the metro area, etc.). For example, a financial institution may utilize the PAR score and/or the PAR report as an additional perspective to its underwriting process on a given building. An existing building owner may leverage the PAR perspective to assess how its building may be perceived relative to comparable buildings along a radius and/or directly update various building attributes to ensure the most accurate information informs the way a given building is evaluated. The prospective tenant may leverage the rating as a point of view in considering options for relocation. A property manager may leverage the rating to understand what makes surrounding buildings rate differently. A potential investor may leverage the rating to inform relative acquisition attractiveness. The property attractiveness rating may be intended to replace existing processes that various types of entities in the real estate value chain may follow. It provides a means by which various entities can consume a distilled perspective on an individual building as derived from considering a large number of factors, including property information, for example.
During a process of system evaluating with the system evaluation model 430, based on the new or updated property data aggregated or received by the PAR system, new derived property attractiveness ratings and newly selected models may be created, updated and or added to the proprietary data store 110 for future retrieval, as illustrated in
In some embodiments, as illustrated in
Below is an example of an API call used for a request to get the property details served by the PAR data layer 220 to the property details module 310.
The API call described above may retrieve relevant information back to the end user's device to be processed for display by the property details module 310. It may include a unique identifier (ID) of the property as stored in the proprietary data store 110.
Below is another example of an API call used for a request to get the property attractiveness rating served by the PAR data layer 220 to the PAR details module 320.
Using the API call described above, a relevant property attractiveness rating may be retrieved back to the end user's device 130 to be processed for display by the PAR details module 320 which overlays data processed by the property details module 310.
Below is another example of an API call used for a request to update information on a property by an expert professional, such as a researcher or an appraiser, to the PAR data layer 220. This may imply a user interacting with the advanced visualization layer 230 of the PAR system 101 to enter certain information about a property.
The API call described above may update certain information about the property back to the proprietary data store 110 which is then picked up by the data processing module 410 and may have an impact in the property attractiveness rating of the particular property.
Below is an example of an API call used for a request to update information on a property via a crowdsourced mechanism to the PAR data layer 220. This may imply a user interacting with the advanced visualization layer 230 to enter certain information about a property.
The API call described above may update certain information about the property back to the proprietary data store 110. The items in bold show the type of updates a non-expert user may make. These updates may need an additional verification step before they are committed as ‘valid’ updates and then passed onto the data processing module 410 and may have an impact in the property attractiveness rating (PAR) of the particular property.
As illustrated in
In some embodiments, the PAR data layer 220 may search for at least one property received via geographical map-based interactive user interface. The PAR data layer 220 may search the database 110 for the property address 808 entered by the end user and process the searched property data to derive a static property attractiveness rating score 810 with no additional input in real time. The PAR data layer 220 may generate a property rating report 812 including the property attractiveness rating score 810 with the corresponding property information. A property rating report 812 may be present to the end user 802 on the geographical map-based interactive user interface. The PAR system 101 may receive additional user inputs 820 such as additional observations and/or attributes from a user 818. The PAR system 101 may generate and update a property rating report 812 based on the additional inputs 820 received from a user 818.
In some embodiment, the PAR data layer 220 may search the database 110 for the property address 808 entered by the end user 802. Meanwhile, the PAR system 101 may receive additional inputs 822 from a user 818, such as live appraisal, E-signatures, report/contract, notifications, other additional observations and attributes, etc. The PAR system 101 may process the searched property data and additional inputs 822 from the user 818 to derive the property attractiveness rating scores 814 which may change based on the user inputs 822. The PAR data layer 220 may generate multiple property attractiveness rating scores 816 based on score templates. Further, the PAR system 101 may generate and update a property rating report 812 based on the property attractiveness rating scores 814 and the additional inputs 822 from a user 818. The system evaluation module 430 of the PAR data layer 220 may update the property rating report 812 in the database 110. The advanced visualization layer 230 may present the property rating report 812 to the end user 802 on the geographical map-based interactive user interface.
Processor(s) 902 may use any known processor technology, including but not limited to graphics processors and multi-core processors. Suitable processors for the execution of a program of instructions may include, by way of example, both general and special purpose microprocessors, and the sole processor or one of multiple processors or cores, of any kind of computer. Generally, a processor may receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer may include a processor for executing instructions and one or more memories for storing instructions and data. Generally, a computer may also include, or be operatively coupled to communicate with, one or more non-transitory computer-readable storage devices for storing data files; such devices include magnetic disks, such as internal hard disks and removable disks; magneto-optical disks; and optical disks. Storage devices suitable for tangibly embodying computer program instructions and data may include all forms of non-volatile memory, including by way of example semiconductor memory devices, such as EPROM, EEPROM, and flash memory devices; magnetic disks such as internal hard disks and removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory may be supplemented by, or incorporated in, ASICs (application-specific integrated circuits).
Input device 904 may be any known input device technology, including but not limited to a pen input, keyboard (including a virtual keyboard), mouse, track ball, and touch-sensitive pad or display. To provide for interaction with a user, the features and functional operations described in the disclosed embodiments may be implemented on a computer having a display device 906 such as a CRT (cathode ray tube) or LCD (liquid crystal display) monitor for displaying information to the user and a keyboard and a pointing device such as a mouse or a trackball by which the user can provide input to the computer. Display device 906 may be any known display technology, including but not limited to display devices using Liquid Crystal Display (LCD) or Light Emitting Diode (LED) technology.
Communication interfaces 908 may be configured to enable the computing device 900 to communicate with other computing or network device across a network, such as via a wired connection, a wireless connection, or a combination of wired and wireless connections. For example, communication interfaces 908 may include an Ethernet interface, an optical interface, a coaxial interface, an infrared interface, a radio frequency (RF) interface, a universal serial bus (USB) interface, a Wi-Fi interface, a cellular network interface, or the like.
Memory 910 may be any computer-readable medium that participates in providing computer program instructions and data to processor(s) 902 for execution, including without limitation, non-volatile storage media (e.g., optical disks, magnetic disks, flash drives, etc.), or volatile storage media (e.g., SDRAM, ROM, etc.). Memory 910 may include various non-transitory computer-readable instructions for implementing an operating system 912 (e.g., Mac OS®, Windows®, Linux), network communication 914, and Application(s) and program modules 916, etc. One program module 916 may be the Property Attractiveness Rating (PAR) application/system 101 shown in
Network communications 914 may include instructions executed to establish and maintain network connections (e.g., software applications for implementing communication protocols, such as TCP/IP, HTTP, Ethernet, telephony, etc.).
Application(s) and program modules 916 may include software application(s) and different functional program modules which are executed by processor(s) 902 to implement the processes described herein and/or other processes. The program modules may include but are not limited to software programs, objects, components, learning models, and data structures that are configured to perform particular tasks or implement particular data types. The processes described herein may also be implemented in operating system 912.
Communication between various network and computing devices may be facilitated by one or more application programming interfaces (APIs). The API may be implemented as one or more calls in program code that send or receive one or more parameters through a parameter list or other structure based on a call convention defined in an API specification document. A parameter may be a constant, a key, a data structure, an object, an object class, a variable, a data type, a pointer, an array, a list, or another call.
The features and functional operations described in the disclosed embodiments may be implemented in one or more computer programs that may be executable on a programmable system including at least one programmable processor coupled to receive data and instructions from, and to transmit data and instructions to, a data storage system, at least one input device, and at least one output device. A computer program is a set of instructions that can be used, directly or indirectly, in a computer to perform a certain activity or bring about a certain result. A computer program may be written in any form of programming language (e.g., Objective-C, Java), including compiled or interpreted languages, and it may be deployed in any form, including as a stand-alone program or as a module, component, subroutine, or other unit suitable for use in a computing environment.
The described features and functional operations described in the disclosed embodiments may be implemented in a computer system that includes a back-end component, such as a data server, or that includes a middleware component, such as an application server or an Internet server, or that includes a front-end component, such as a user computer having a graphical user interface or an Internet browser, or any combination thereof. The components of the system may be connected by any form or medium of digital data communication such as a communication network. Examples of communication networks include, e.g., a telephone network, a LAN, a WAN, and the computers and networks forming the Internet.
The computer system may include user computing devices and application servers. A user or client computing device and server may generally be remote from each other and may typically interact through a network. The relationship of client computing devices and server may arise by virtue of computer programs running on the respective computers and having a client-server relationship to each other.
While various embodiments have been described above, it should be understood that they have been presented by way of example and not limitation. It will be apparent to persons skilled in the relevant art(s) that various changes in form and detail can be made therein without departing from the spirit and scope. In fact, after reading the above description, it will be apparent to one skilled in the relevant art(s) how to implement alternative embodiments. For example, other steps may be provided, or steps may be eliminated, from the described flows, and other components may be added to, or removed from, the described systems. Accordingly, other implementations are within the scope of the following claims.
In addition, it should be understood that any figures which highlight the functionality and advantages are presented for example purposes only. The disclosed methodology and system are each sufficiently flexible and configurable such that they may be utilized in ways other than that shown.
