Method, apparatus, and computer program product for predicting web browsing behaviors of consumers
Embodiments of the present invention provide methods, systems, apparatuses, and computer program products for predicting consumer behavior. In one embodiment a method is provided comprising automatically creating a link graph comprising nodes representing webpages, links representing hyperlinks, and weights for each link representing a number of times a hyperlink associated with the respective link redirected the a user devices from a webpage associated with a first node connected to the link to a webpage associated with a second node connected to the link; and determining based on the link graph a programmatically expected path for a particular user, wherein the programmatically expected path identifies, at least, two or more webpages that the particular user is programmatically expected to visit and specifying a programmatically expected order at which the particular user will visit the webpages.
1. An apparatus comprising at least one processor and at least one non-transitory memory comprising program code, wherein the at least one non-transitory memory and the program code are configured to, with the at least one processor, cause the apparatus to:
generate a link graph for a website comprising a plurality of webpages, the link graph comprising a plurality of nodes and a plurality of links, wherein a node of the plurality of nodes represents a webpage of the plurality of webpages and a link of the plurality of links represents a hyperlink of one or more hyperlinks;
modify the link graph by eliminating those links of the plurality of links from the link graph that are associated with a desired computational expense below a first threshold or a desired accuracy below a second threshold; and
generate a programmatically expected path for a particular user through the link graph based at least in part on a number of times each link associated with a first node was accessed.
2. The apparatus of claim 1 , wherein the link graph is generated based at least in part on data comprising the one or more hyperlinks associated with the plurality of webpages.
3. The apparatus of claim 2 , wherein the link graph is further generated based at least in part on historical data associated with the plurality of webpages.
4. The apparatus of claim 1 , wherein each link is associated with a weight value.
5. The apparatus of claim 4 , wherein the weight value represents a number of times a hyperlink represented by the link was accessed by a user device.
6. The apparatus of claim 5 , wherein the weight value represents a normalized count of the number of times a hyperlink represented by the link was accessed by a user device.
7. The apparatus of claim 1 , wherein generating the programmatically expected path for the particular user is further based at least in part on historical data, retrieved from a database, for the particular user of the website.
8. The apparatus of claim 4 , wherein generating the programmatically expected path for the particular user comprises:
calculating a transition probability based on a ratio of weight values between the first node, a second node, and a sum of all weight values for transitions from the first node to any other node of the link graph.
9. The apparatus of claim 8 , wherein generating the programmatically expected path for the particular user is further based at least in part on the transition probability.
10. The apparatus of claim 8 , wherein the transition probability between the first node and the second node represents a likelihood that the particular user will be redirected from a webpage represented by the first node to a webpage represented by the second node.
11. The apparatus of claim 4 , wherein a prediction model configured to generate one or more predictions associated with the website is trained based at least in part on one or more of: (i) the weight value associated with each link, (ii) the link graph, and (iii) historical data.
12. A computer-implemented method comprising:
generating a link graph for a website comprising a plurality of webpages, the link graph comprising a plurality of nodes and a plurality of links, wherein a node of the plurality of nodes represents a webpage of the plurality of webpages and a link of the plurality of links represents a hyperlink of one or more hyperlinks;
modifying the link graph by eliminating those links of the plurality of links from the link graph that are associated with a desired computational expense below a first threshold or a desired accuracy below a second threshold; and
generating a programmatically expected path for a particular user through the link graph based at least in part on a number of times each link associated with a first node was accessed.
13. The computer-implemented method of claim 12 , wherein the link graph is generated based at least in part on data comprising the one or more hyperlinks associated with the plurality of webpages.
14. The computer-implemented method of claim 13 , wherein the link graph is further generated based at least in part on historical data associated with the plurality of webpages.
15. The computer-implemented method of claim 12 , wherein each link is associated with a weight value.
16. The computer-implemented method of claim 15 , wherein the weight value represents a number of times a hyperlink represented by the link was accessed by a user device.
17. The computer-implemented method of claim 16 , wherein the weight value represents a normalized count of the number of times a hyperlink represented by the link was accessed by a user device.
18. The computer-implemented method of claim 12 , wherein generating the programmatically expected path for the particular user is further based at least in part on historical data, retrieved from a database, for the particular user of the website.
19. The computer-implemented method of claim 15 , wherein generating the programmatically expected path for the particular user comprises:
calculating a transition probability based on a ratio of weight values between the first node, a second node, and a sum of all weight values for transitions from the first node to any other node of the link graph; and
generating the programmatically expected path for the particular user is further based at least in part on the transition probability.
20. A computer program product, the computer program product comprising at least one non-transitory computer-readable storage medium having computer-readable program code portions stored therein, the computer-readable program code portions configured to:
generate a link graph for a website comprising a plurality of webpages, the link graph comprising a plurality of nodes and a plurality of links, wherein a node of the plurality of nodes represents a webpage of the plurality of webpages and a link of the plurality of links represents a hyperlink of one or more hyperlinks;
modify the link graph by eliminating those links of the plurality of links from the link graph that are associated with a desired computational expense below a first threshold or a desired accuracy below a second threshold; and
generate a programmatically expected path for a particular user through the link graph based at least in part on a number of times each link associated with a first node was accessed.