Collaborative filtering to generate recommendations
A system, method, and computer-readable medium are disclosed for performing a recommendation operation, comprising: optimizing a product list to provide an optimized product list for use when generating a recommendation for an account; optimizing a neighbor set to provide an optimized neighbor set for use when generating the recommendation for the account; boosting a self-cosine similarity metric to provide a boosted self-cosine similarity metric, the self-cosine similarity metric corresponding to the account; and, providing a recommendation for the account, the recommendation being based on the optimized product list, the optimized neighbor set and the boosted self-cosine similarity metric.
1. A computer-implementable method for performing a recommendation operation, comprising:
optimizing a product list to provide an optimized product list for use when generating a recommendation for an account, the optimized product list comprising a plurality of information handling systems, each of the plurality of information handling systems comprising a plurality of components;
optimizing a neighbor set to provide an optimized neighbor set for use when generating the recommendation for the account, the neighbor set representing interactions of other users from a same type of business;
boosting a self-cosine similarity metric to provide a boosted self-cosine similarity metric, the self-cosine similarity metric corresponding to the account, the boosted self-cosine metric being based upon past purchases of the account, the boosted self-cosine metric using a homogeneity factor of the account;
providing a recommendation for components to include in a product for the account, the recommendation being based on the optimized product list, the optimized neighbor set and the boosted self-cosine similarity metric;
fabricating the product via a custom product fabrication system, the product being fabricated to include components identified via the recommendation;
generating the homogeneity factor for the account as:
Homogeneity Factor (HF)=(Total number of Line of Businesses (LOBs))÷(Number of LOBs an account has purchased in the past)
generating a revenue proportion value, the revenue proportion value representing revenue of the account by business segment, the revenue proportion value being calculated as:
Revenue proportion (RP)=(AR for a LOB+PP for a LOB)÷(AR for all LOBs+PP for all LOBs)
For each account:
AR=Per quarter average revenue for each LOB
PP=Weighted Open Pipeline for immediate next quarter for each LOB
generating a rank threshold derived from each neighbor set, the rank threshold being calculated as:
Rank Threshold (RT)=1÷(Lowest Account Coverage (%))
where RT is the number of accounts in the neighbor set;
generating a single dimension representation of lines of business having a revenue proportion for the account, the single dimension representation of lines of business being calculated as:
Single dimension representation of LOB RPs (TRP)=Transpose (RP Matrix) m by n *(PC1 Loadings) n by 1 ;
generating a weighted average score, the weighted average score using a boosted self-cosine similarity metric for the account and boosted self-cosine similarity metrics for each neighbor account; and,
using the weighted average score to learn recommendations from the account and the neighbor accounts.
2. The method of claim 1 , further comprising:
normalizing the boosted self-cosine metric, the normalizing providing an adjusted boosted self-cosine metric.
3. The method of claim 1 , wherein:
the recommendation operation uses collaborative filtering with self-boosted and principal component adjusted weights using an optimal neighbor set to generate recommendations.
4. The method of claim 1 , wherein:
the recommendation operation is tailored for business to business (B2B) transactions.
5. The method of claim 1 , wherein:
an interaction of the account is normalized via a principal component approach so that the recommendation operation learns from neighbors of the account.
6. The method of claim 1 , wherein:
the recommendation operation considers an open sales pipeline when generating the recommendation.
7. A system comprising:
a processor;
a data bus coupled to the processor; and
a non-transitory, computer-readable storage medium embodying computer program code, the non-transitory, computer-readable storage medium being coupled to the data bus, the computer program code interacting with a plurality of computer operations and comprising instructions executable by the processor and configured for:
optimizing a product list to provide an optimized product list for use when generating a recommendation for an account, the optimized product list comprising a plurality of information handling systems, each of the plurality of information handling systems comprising a plurality of components;
optimizing a neighbor set to provide an optimized neighbor set for use when generating the recommendation for the account, the neighbor set representing interactions of other users from a same type of business;
boosting a self-cosine similarity metric to provide a boosted self-cosine similarity metric, the self-cosine similarity metric corresponding to the account, the boosted self-cosine metric being based upon past purchases of the account the boosted self-cosine metric taking into account a homogeneity factor of the account;
providing a recommendation for components to include in a product for the account, the recommendation being based on the optimized product list, the optimized neighbor set and the boosted self-cosine similarity metric;
fabricating the product via a custom product fabrication system, the product being fabricated to include components identified via the recommendation;
generating the homogeneity factor for the account as:
Homogeneity Factor (HF)=(Total number of Line of Businesses (LOBs))÷(Number of LOBs an account has purchased in the past)
generating a revenue proportion value, the revenue proportion value representing revenue of the account by business segment, the revenue proportion value being calculated as:
Revenue proportion (RP)=(AR for a LOB+PP for a LOB)÷(AR for all LOBs+PP for all LOBs)
For each account:
AR=Per quarter average revenue for each LOB
PP=Weighted Open Pipeline for immediate next quarter for each LOB
generating a rank threshold derived from each neighbor set, the rank threshold being calculated as:
Rank Threshold (RT)=1÷(Lowest Account Coverage (%))
where RT is the number of accounts in the neighbor set;
generating a single dimension representation of lines of business having a revenue proportion for the account, the single dimension representation of lines of business being calculated as:
Single dimension representation of LOB RPs (TRP)=Transpose (RP Matrix) m by n *(PC1 Loadings) n by 1 ;
generating a weighted average score, the weighted average score using a boosted self-cosine similarity metric for the account and boosted self-cosine similarity metrics for each neighbor account; and,
using the weighted average score to learn recommendations from the account and the neighbor accounts.
8. The system of claim 7 , wherein the instructions executable by the processor are further configured for:
normalizing the boosted self-cosine metric, the normalizing providing an adjusted boosted self-cosine metric.
9. The system of claim 7 , wherein:
the recommendation operation uses collaborative filtering with self-boosted and principal component adjusted weights using an optimal neighbor set to generate recommendations.
10. The system of claim 7 , wherein:
the recommendation operation is tailored for business to business (B2B) transactions.
11. The system of claim 7 , wherein:
an interaction of the account is normalized via a principal component approach so that the recommendation operation learns from neighbors of the account.
12. The system of claim 7 , wherein:
the recommendation operation considers an open sales pipeline when generating the recommendation.
13. A non-transitory, computer-readable storage medium embodying computer program code, the computer program code comprising computer executable instructions configured for:
optimizing a product list to provide an optimized product list for use when generating a recommendation for an account, the optimized product list comprising a plurality of information handling systems, each of the plurality of information handling systems comprising a plurality of components;
optimizing a neighbor set to provide an optimized neighbor set for use when generating the recommendation for the account, the neighbor set representing interactions of other users from a same type of business;
boosting a self-cosine similarity metric to provide a boosted self-cosine similarity metric, the self-cosine similarity metric corresponding to the account, the boosted self-cosine metric being based upon past purchases of the account the boosted self-cosine metric taking into account a homogeneity factor of the account;
providing a recommendation for components to include in a product for the account, the recommendation being based on the optimized product list, the optimized neighbor set and the boosted self-cosine similarity metric;
fabricating the product via a custom product fabrication system, the product being fabricated to include components identified via the recommendation;
generating the homogeneity factor for the account as:
Homogeneity Factor (HF)=(Total number of Line of Businesses (LOBs))÷(Number of LOBs an account has purchased in the past)
generating a revenue proportion value, the revenue proportion value representing revenue of the account by business segment, the revenue proportion value being calculated as:
Revenue proportion (RP)=(AR for a LOB+PP for a LOB)÷(AR for all LOBs+PP for all LOBs)
For each account:
AR=Per quarter average revenue for each LOB
PP=Weighted Open Pipeline for immediate next quarter for each LOB
generating a rank threshold derived from each neighbor set, the rank threshold being calculated as:
Rank Threshold (RT)=1÷(Lowest Account Coverage (%))
where RT is the number of accounts in the neighbor set;
generating a single dimension representation of lines of business having a revenue proportion for the account, the single dimension representation of lines of business being calculated as:
Single dimension representation of LOB RPs (TRP)=Transpose (RP Matrix) m by n *(PC1 Loadings) n by 1 ;
generating a weighted average score, the weighted average score using a boosted self-cosine similarity metric for the account and boosted self-cosine similarity metrics for each neighbor account; and,
using the weighted average score to learn recommendations from the account and the neighbor accounts.
14. The non-transitory, computer-readable storage medium of claim 13 , wherein the computer executable instructions are further configured for:
normalizing the boosted self-cosine metric, the normalizing providing an adjusted boosted self-cosine metric.
15. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the recommendation operation uses collaborative filtering with self-boosted and principal component adjusted weights using an optimal neighbor set to generate recommendations.
16. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the recommendation operation is tailored for business to business (B2B) transactions.
17. The non-transitory, computer-readable storage medium of claim 13 , wherein:
an interaction of the account is normalized via a principal component approach so that the recommendation operation learns from neighbors of the account.
18. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the recommendation operation considers an open sales pipeline when generating the recommendation.
19. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the computer executable instructions are deployable to a client system from a server system at a remote location.
20. The non-transitory, computer-readable storage medium of claim 13 , wherein:
the computer executable instructions are provided by a service provider to a user on an on-demand basis.