same reccomendations for the most of the users

[kinderp]

Hi, i'm trying to adapt movielens example with my data.
My goal is using dsstne to evaluate my reccomendations engine
but i think should be some problem because i get the same reccomendations
for the most of the user.

I'm using a csv with the same ratings format of movielens csv

user_id, item_id, rating, timestamp
1,22611,5,0
0,26475,5,0
6,26475,1,0
8,26475,5,0
11,26475,5,0
16,26475,4,0
0,26476,5,0
3,26476,1,0
4,26476,1,0
0,6537,5,0

I followed this guide to generate predictions with my data.

This is an head of my recs file

0	355367,0.141:125965,0.096:125944,0.048:353927,0.040:198763,0.040:279172,0.040:198798,0.040:349333,0.040:353999,0.040:198884,0.040:
8	355367,0.141:125965,0.048:125944,0.048:353927,0.040:198763,0.040:279172,0.040:198798,0.040:349333,0.040:353999,0.040:198884,0.040:
11	125965,0.047:125944,0.047:353927,0.039:198763,0.039:279172,0.039:198798,0.039:349333,0.039:353999,0.039:198884,0.039:316070,0.039:
16	125965,0.047:125944,0.047:353927,0.039:198763,0.039:279172,0.039:198798,0.039:349333,0.039:353999,0.039:198884,0.039:316070,0.039:
4	355367,0.141:125965,0.048:125944,0.048:353927,0.040:198763,0.040:279172,0.040:198798,0.040:349333,0.040:353999,0.040:198884,0.040:
3	355367,0.141:125965,0.048:353927,0.040:198763,0.040:279172,0.040:198798,0.040:349333,0.040:353999,0.040:198884,0.040:316070,0.040:
6	355367,0.141:125965,0.048:125944,0.048:353927,0.040:198763,0.040:279172,0.040:198798,0.040:349333,0.040:353999,0.040:198884,0.040:
13	125965,0.047:125944,0.047:353927,0.039:198763,0.039:279172,0.039:198798,0.039:349333,0.039:353999,0.039:198884,0.039:316070,0.039:
14	125965,0.047:125944,0.047:353927,0.039:198763,0.039:279172,0.039:198798,0.039:349333,0.039:353999,0.039:198884,0.039:316070,0.039:
18	125965,0.047:125944,0.047:353927,0.039:198763,0.039:279172,0.039:198798,0.039:349333,0.039:353999,0.039:198884,0.039:316070,0.039:

this is my config.json

{
"Version" : 0.8,
"Name" : "AIV NNC",
"Kind" : "FeedForward",  
"ShuffleIndices" : false,
"ScaledMarginalCrossEntropy" : {
"oneTarget" : 1.0,
"zeroTarget" : 0.0,
"oneScale" : 1.0,
"zeroScale" : 1.0
},
"Layers" : [
{ "Name" : "Input", "Kind" : "Input", "N" : "auto", "DataSet" : "just_gl_input", "Sparse" : true }, 
{ "Name" : "Hidden1", "Kind" : "Hidden", "Type" : "FullyConnected", "N" : 1536, "Activation" : "Relu", "Sparse" : false, "pDropout" : 0.37, "WeightInit" : { "Scheme" : "Gaussian", "Scale" : 0.01 } },
{ "Name" : "Hidden2", "Kind" : "Hidden", "Type" : "FullyConnected", "N" : 1536, "Activation" : "Relu", "Sparse" : false, "pDropout" : 0.37, "WeightInit" : { "Scheme" : "Gaussian", "Scale" : 0.01 } },  
{ "Name" : "Hidden3", "Kind" : "Hidden", "Type" : "FullyConnected", "N" : 1536, "Activation" : "Relu", "Sparse" : false, "pDropout" : 0.37, "WeightInit" : { "Scheme" : "Gaussian", "Scale" : 0.01 } },  
{ "Name" : "Output", "Kind" : "Output", "Type" : "FullyConnected",  "DataSet" : "just_gl_output", "N" : "auto", "Activation" : "Sigmoid", "Sparse" : true , "WeightInit" : { "Scheme" : "Gaussian", "Scale" : 0.01, "Bias" : -10.2 }}
],
"ErrorFunction" : "ScaledMarginalCrossEntropy"
}

Any advices or ideas are welcome,
Thanks in advance

[spacelover1]

Hi Antonio (@kinderp ),

I need to evaluate dsstne topN recommendations on my own dataset, I was wondering if you could find any ways to do this finally, if you were able to do the evaluation, would you please share your solutions?

Best,