mlpack_cf(1)

1mlpack_cf(1)                General Commands Manual               mlpack_cf(1)
2
3
4

NAME

6       mlpack_cf - collaborating filtering
7

SYNOPSIS

9        mlpack_cf [-h] [-v]
10

DESCRIPTION

12       This  program  performs  collaborative  filtering  (CF)  on  the  given
13       dataset. Given a list of user, item and  preferences  (--training_file)
14       the  program will perform a matrix decomposition and then can perform a
15       series of actions related to collaborative filtering. Alternately,  the
16       program can load an existing saved CF model with the --input_model_file
17       (-m) option and then use that model to provide recommendations or  pre‐
18       dict values.
19
20       The  input  file should contain a 3-column matrix of ratings, where the
21       first column is the user, the second column is the item, and the  third
22       column  is  that  user's  rating of that item. Both the users and items
23       should be numeric indices, not names. The indices are assumed to  start
24       from 0.
25
26       A  set of query users for which recommendations can be generated may be
27       specified with the --query_file (-q) option;  alternately,  recommenda‐
28       tions  may be generated for every user in the dataset by specifying the
29       --all_user_recommendations (-A) option. In addition, the number of rec‐
30       ommendations per user to generate can be specified with the --recommen‐
31       dations (-r) parameter, and the number of similar users  (the  size  of
32       the  neighborhood) to be considered when generating recommendations can
33       be specified with the --neighborhood (-n) option.
34
35       For performing the matrix  decomposition,  the  following  optimization
36       algorithms  can  be  specified  via  the  --algorithm  (-a)  parameter:
37       ’RegSVD' -- Regularized SVD using a SGD optimizer ’NMF' -- Non-negative
38       matrix  factorization  with  alternating  least  squares  update  rules
39       ’BatchSVD' -- SVD  batch  learning  ’SVDIncompleteIncremental'  --  SVD
40       incomplete  incremental  learning  ’SVDCompleteIncremental' -- SVD com‐
41       plete incremental learning
42
43       A trained model may be saved to a  file  with  the  --output_model_file
44       (-M) parameter.
45

OPTIONAL INPUT OPTIONS

47       --algorithm (-a) [string]
48              Algorithm  used  for matrix factorization.  Default value 'NMF'.
49              --all_user_recommendations (-A) Generate recommendations for all
50              users.
51
52       --help (-h)
53              Default help info.
54
55       --info [string]
56              Get  help  on  a  specific  module or option.  Default value ''.
57              --input_model_file (-m) [string] File to load trained  CF  model
58              from.  Default value ''.  --iteration_only_termination (-I) Ter‐
59              minate only when the maximum number of iterations is reached.
60
61       --max_iterations (-N) [int]
62              Maximum number of iterations. Default value
63
64              1000.
65
66
67       --min_residue (-r) [double]
68              Residue required to terminate the  factorization  (lower  values
69              generally mean better fits).  Default value 1e-05.
70
71       --neighborhood (-n) [int]
72              Size  of  the neighborhood of similar users to consider for each
73              query user. Default value 5.
74
75       --query_file (-q) [string]
76              List of users for which recommendations  are  to  be  generated.
77              Default value ''.
78
79       --rank (-R) [int]
80              Rank  of decomposed matrices (if 0, a heuristic is used to esti‐
81              mate the rank). Default value
82
83              0.
84
85                  --recommendations (-c) [int] Number  of  recommendations  to
86                  generate for each query user. Default value 5.
87
88       --seed (-s) [int]
89              Set the random seed (0 uses std::time(NULL)).  Default value 0.
90
91       --test_file (-T) [string]
92              Test  set  to  calculate  RMSE  on.  Default value ’'.  --train‐
93              ing_file (-t) [string] Input dataset to perform CF  on.  Default
94              value ’'.
95
96       --verbose (-v)
97              Display  informational  messages and the full list of parameters
98              and timers at the end of execution.
99
100       --version (-V)
101              Display the version of mlpack.
102

OPTIONAL OUTPUT OPTIONS

104       --output_file (-o) [string]
105              File to  save  output  recommendations  to.  Default  value  ''.
106              --output_model_file  (-M) [string] File to save trained CF model
107              to. Default value ’'.
108

ADDITIONAL INFORMATION

111       For further information, including relevant papers, citations, and the‐
112       ory, For further information, including relevant papers, citations, and
113       theory, consult the documentation  found  at  http://www.mlpack.org  or
114       included    with    your    consult    the   documentation   found   at
115       http://www.mlpack.org or included with  your  DISTRIBUTION  OF  MLPACK.
116       DISTRIBUTION OF MLPACK.
117
118
119
120                                                                  mlpack_cf(1)