Data-driven classifier

1. A method comprising:

receiving a transaction dataset including data representative of a plurality of transactions, data representative of each of the plurality of transactions including at least one product purchased within the transaction;

processing the transaction dataset according to Word2vec algorithm to generate a data representation for each of a plurality of unique products included in the transaction dataset, each generated data representation represents a context of a unique product with regard to each of a plurality of other products included in the generated data representation, the processing including:

treating the plurality of unique products as words and consumer purchase transactions as sentences;

mapping, with the data contextualizing algorithm, each unique word of the treated plurality of unique products and consumer purchase transactions to a vector within a vector space of N dimensions, the vector representing the context of the unique product and an associated consumer purchase transaction within the vector space, wherein N is an integer greater than 1; and

wherein distances between pairs of vectors within the vector space are representative of similarity levels such that a lesser distance represents a greater similarity and a greater distance represents a lesser similarity;

processing the generated data representations of the plurality of unique products included in the transaction dataset according to a clustering algorithm to partition the plurality of unique products into a number of product clusters, each product cluster containing respective products that are similar to one another; and

storing and outputting a cluster representation of each the product clusters including data identifying products and a product cluster to which the identified products are partitioned, the cluster representation including at least one of complimentary products and substitute products.

2. The method of claim 1 , wherein the data contextualizing algorithm is performed by a context module that executes on a computing device that receives the transaction dataset electronically.

3. The method of claim 2 , wherein the clustering algorithm is performed by a clustering module that executes on the computing device.

4. The method of claim 3 , wherein the clustering module performs k-means algorithm against the data representations outputted by the context module.

5. The method of claim 1 , wherein the number of product clusters output by the clustering algorithm is based on a configuration setting.

6. The method of claim 1 , wherein cluster representations of the product clusters are accessed by other processes that identify one or both of interchangeable and complimentary products.

7. A non-transitory computer-readable medium storing computer-executable instructions thereon to implement a method comprising:

receiving a transaction dataset including data representative of a plurality of transactions, data representative of each of the plurality of transactions including at least one product purchased within the transaction;

processing the transaction dataset according to Word2vec algorithm to generate a data representation for each of a plurality of unique products included in the transaction dataset, each generated data representation represents a context of a unique product with regard to each of a plurality of other products included in the generated data representation, the processing including:

treating the plurality of unique products as words and consumer purchase transactions as sentences;

mapping, with the data contextualizing algorithm, each unique word of the treated plurality of unique products and consumer purchase transactions to a vector within a vector space of N dimensions, the vector representing the context of the unique product and an associated consumer purchase transaction within the vector space, wherein Nis an integer greater than 1; and

wherein distances between pairs of vectors within the vector space are representative of similarity levels such that lesser distance represents a greater similarity and a greater distance represents a lesser similarity;

processing the generated data representations of the plurality of unique products included in the transaction dataset according to a clustering algorithm to partition the plurality of unique products into a number of product clusters, each product cluster containing respective products that are similar to one another; and

storing and outputting a cluster representation of each of the product clusters including data identifying products and a product cluster to which the identified products are partitioned, the cluster representation including at least one of complimentary products and substitute products.

8. The non-transitory computer readable medium of claim 7 , wherein the data contextualizing algorithm is performed by a context module that executes on a computing device that either receives or retrieves the transaction dataset electronically.

9. The non-transitory computer readable medium of claim 8 , wherein the clustering algorithm is performed by a clustering module that executes on the computing device.

10. The non-transitory computer readable medium of claim 9 , wherein the clustering module performs k-means algorithm against the data representations outputted by the context module.

11. A system comprising:

a processor and a memory device storing instructions executable by the processor to perform a method comprising:

receiving a transaction dataset including data representative of a plurality of transactions, data representative of each of the plurality of transactions including at least one product purchased within the transaction;

processing the transaction dataset according to Word2vec algorithm to generate a data representation for each of at a plurality of unique products included in the transaction dataset, each generated data representation represents a context of a unique product with regard to each of a plurality of other products included in the generated data representation, the processing including:

treating the plurality of unique products as words and consumer purchase transactions as sentences;

mapping, with the data contextualizing algorithm, each unique word of the treated plurality of unique products and consumer purchase transactions to a vector within a vector space of N dimensions, the vector representing the context of the unique product and an associated consumer purchase transaction within the vector spaee, wherein N is an integer greater than 1; and

wherein distances between pairs of vectors within the vector space are representative of similarity levels such that a lesser distacne represents a greater similarity and a greater distance represents a lesser similarity;

processing the generated data representations of the plurality of unique products included in the transaction dataset according to a clustering algorithm to partition the plurality of unique products into a number of product clusters, each product duster containing respective products that are similar to one another; and

storing and outputting a duster representation of each of the product dusters including data identifying products and a product cluster to which the identified products are partitioned, the duster representation including at least one of complimentary products and substitute products.

12. The method of claim 11 , wherein:

the clustering algorithm is executed by a clustering module that executes of the computing device; and

the clustering module performs k-means algorithm against the data representations outputted by the context module.