{
  "total": 429,
  "topics": [
    {
      "id": 40232,
      "title": "Welcome to the Porto Seguro competition!",
      "comment_count": 71,
      "views": 0,
      "votes": 56
    },
    {
      "id": 40222,
      "title": "Welcome!",
      "comment_count": 53,
      "views": 0,
      "votes": 47
    },
    {
      "id": 44629,
      "title": "1st place with representation learning",
      "comment_count": 463,
      "views": 0,
      "votes": 1098
    },
    {
      "id": 43976,
      "title": "About (hard) work and (soft) expertise",
      "comment_count": 30,
      "views": 0,
      "votes": 52
    },
    {
      "id": 44608,
      "title": "3rd place solution",
      "comment_count": 33,
      "views": 0,
      "votes": 124
    },
    {
      "id": 44002,
      "title": "The biggest competition on Kaggle.com!!!",
      "comment_count": 18,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43306,
      "title": "If you are depressed",
      "comment_count": 47,
      "views": 0,
      "votes": 25
    },
    {
      "id": 180961,
      "title": "A simplified approach - Predicting Insurance Claim",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44092,
      "title": "Are red cars more accident prone?",
      "comment_count": 3,
      "views": 0,
      "votes": 8
    },
    {
      "id": 40577,
      "title": "Random Forest Classifier",
      "comment_count": 20,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40848,
      "title": "Encoding of categorical variables",
      "comment_count": 13,
      "views": 0,
      "votes": 7
    },
    {
      "id": 83788,
      "title": "Submission error",
      "comment_count": 7,
      "views": 0,
      "votes": 1
    },
    {
      "id": 51410,
      "title": "which normalization methods for DNN?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41029,
      "title": "Just a tip: Gini = 2*AUC - 1",
      "comment_count": 12,
      "views": 0,
      "votes": 10
    },
    {
      "id": 40368,
      "title": "What's the suitable loss function for Gini?",
      "comment_count": 18,
      "views": 0,
      "votes": 15
    },
    {
      "id": 40618,
      "title": "Imbalanced dataset",
      "comment_count": 17,
      "views": 0,
      "votes": 10
    },
    {
      "id": 43254,
      "title": "Really LB score is meaningful? (from recent kernels)",
      "comment_count": 25,
      "views": 0,
      "votes": 8
    },
    {
      "id": 92753,
      "title": "submission error",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41159,
      "title": "XGBoost Python with GPU (CUDA)",
      "comment_count": 16,
      "views": 0,
      "votes": 1
    },
    {
      "id": 83893,
      "title": "Late submissions broken",
      "comment_count": 4,
      "views": 0,
      "votes": 3
    },
    {
      "id": 43394,
      "title": "negative prediction in XGboost",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 43842,
      "title": "LambdaRank in TensorFlow",
      "comment_count": 5,
      "views": 0,
      "votes": 23
    },
    {
      "id": 41489,
      "title": "most important feature: ps_car_13 - car's mileage?",
      "comment_count": 35,
      "views": 0,
      "votes": 49
    },
    {
      "id": 44254,
      "title": "should random seed be a parameter we try to tune",
      "comment_count": 13,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40845,
      "title": "New joiner question - what is the meaning of \" features that belong to similar groupings\" ?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 41018,
      "title": "XGB, LGBM and Catboost ??",
      "comment_count": 23,
      "views": 0,
      "votes": 14
    },
    {
      "id": 44987,
      "title": "Target Encoding in Stacking ",
      "comment_count": 31,
      "views": 0,
      "votes": 9
    },
    {
      "id": 56048,
      "title": "Predicting probability using GINI?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41359,
      "title": "Will scale_pos_weight parameter in xgboost help the imbalance data issue?",
      "comment_count": 15,
      "views": 0,
      "votes": 6
    },
    {
      "id": 51064,
      "title": "Can I publish EDA of this competition?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 44558,
      "title": "三个臭皮匠's approach(s)",
      "comment_count": 68,
      "views": 0,
      "votes": 192
    },
    {
      "id": 44821,
      "title": "23rd place solution",
      "comment_count": 12,
      "views": 0,
      "votes": 22
    },
    {
      "id": 44614,
      "title": "Solution 1178 Public / 29 Private",
      "comment_count": 60,
      "views": 0,
      "votes": 108
    },
    {
      "id": 44586,
      "title": "Solution Sharing ",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 49817,
      "title": "Private LB score very close to CV score",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 48743,
      "title": "Insight gained from the competition ?",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43703,
      "title": "Use VPS to do gridsearch",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42638,
      "title": "What about using WOE and IV ? ",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 47464,
      "title": "Anyone interested in beta testing AutoML on this dataset?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 47321,
      "title": "[Video] Gold medal solution by utility (English subtitles)",
      "comment_count": 0,
      "views": 0,
      "votes": 12
    },
    {
      "id": 44601,
      "title": "Taylor-made NN for 0.285 PLB (part of solution of 8º)",
      "comment_count": 32,
      "views": 0,
      "votes": 57
    },
    {
      "id": 46612,
      "title": "Trying to reproduce the winning solution",
      "comment_count": 4,
      "views": 0,
      "votes": 7
    },
    {
      "id": 43144,
      "title": "Is LB score trustworthy?",
      "comment_count": 63,
      "views": 0,
      "votes": 13
    },
    {
      "id": 44698,
      "title": "Let's reproduce Michael Jahrer's solution in python",
      "comment_count": 3,
      "views": 0,
      "votes": 4
    },
    {
      "id": 44571,
      "title": "Approach(3rd in public and 8th in private)",
      "comment_count": 39,
      "views": 0,
      "votes": 49
    },
    {
      "id": 45893,
      "title": "Can't use PCA",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 44546,
      "title": "Congratulations and Thank You",
      "comment_count": 47,
      "views": 0,
      "votes": 89
    },
    {
      "id": 44579,
      "title": "18th Place Solution - Careful Ensembling + Resampling Diversity",
      "comment_count": 17,
      "views": 0,
      "votes": 60
    },
    {
      "id": 44348,
      "title": "I'm off this competition, alea jacta est !",
      "comment_count": 48,
      "views": 0,
      "votes": 83
    },
    {
      "id": 44803,
      "title": "Inspirational note for beginners",
      "comment_count": 9,
      "views": 0,
      "votes": 21
    },
    {
      "id": 44700,
      "title": "9th place solution",
      "comment_count": 11,
      "views": 0,
      "votes": 40
    },
    {
      "id": 44829,
      "title": "Unfair Bronze Medal",
      "comment_count": 8,
      "views": 0,
      "votes": 22
    },
    {
      "id": 44678,
      "title": "Compiled list of solutions",
      "comment_count": 19,
      "views": 0,
      "votes": 41
    },
    {
      "id": 44853,
      "title": "Public kernels and free lunch",
      "comment_count": 15,
      "views": 0,
      "votes": 13
    },
    {
      "id": 44659,
      "title": "genetic algorithm solution (20th place) - very long read ",
      "comment_count": 24,
      "views": 0,
      "votes": 77
    },
    {
      "id": 44472,
      "title": "New Kagglers to follow in 2018",
      "comment_count": 24,
      "views": 0,
      "votes": 30
    },
    {
      "id": 44638,
      "title": "Porto Seguro is proud of this competition",
      "comment_count": 12,
      "views": 0,
      "votes": 45
    },
    {
      "id": 45109,
      "title": "A few questions on xgboost",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 44548,
      "title": "Best ISngle Model on Private LB?",
      "comment_count": 36,
      "views": 0,
      "votes": 22
    },
    {
      "id": 44784,
      "title": "39th solution utilizing Pr(is_test|X) by adversarial validation",
      "comment_count": 10,
      "views": 0,
      "votes": 24
    },
    {
      "id": 45143,
      "title": "win a silver medal with kernels",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 44759,
      "title": "Keep It Simple (21st solution & *mea culpa* for missing gold)",
      "comment_count": 21,
      "views": 0,
      "votes": 12
    },
    {
      "id": 44711,
      "title": "35th place solution",
      "comment_count": 8,
      "views": 0,
      "votes": 23
    },
    {
      "id": 44655,
      "title": "17th solution - seed selection, NN, ensembling..",
      "comment_count": 15,
      "views": 0,
      "votes": 30
    },
    {
      "id": 44642,
      "title": "12th place solution",
      "comment_count": 9,
      "views": 0,
      "votes": 29
    },
    {
      "id": 44545,
      "title": "31th/5k+ with 1 single model without stacking and blending possible, wow!",
      "comment_count": 21,
      "views": 0,
      "votes": 21
    },
    {
      "id": 44492,
      "title": "I'm stupid",
      "comment_count": 73,
      "views": 0,
      "votes": 27
    },
    {
      "id": 44843,
      "title": "poor LightGBM results",
      "comment_count": 12,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44541,
      "title": "Some reward",
      "comment_count": 38,
      "views": 0,
      "votes": 40
    },
    {
      "id": 44564,
      "title": "Thank you ",
      "comment_count": 10,
      "views": 0,
      "votes": 22
    },
    {
      "id": 44727,
      "title": "Leaderboard validated so quickly",
      "comment_count": 21,
      "views": 0,
      "votes": 13
    },
    {
      "id": 44307,
      "title": "Where is the reward?",
      "comment_count": 18,
      "views": 0,
      "votes": 48
    },
    {
      "id": 44606,
      "title": "Why was ensembling difficult?",
      "comment_count": 7,
      "views": 0,
      "votes": 10
    },
    {
      "id": 44429,
      "title": "Congratz @Michael and @Bojan",
      "comment_count": 25,
      "views": 0,
      "votes": 26
    },
    {
      "id": 44651,
      "title": "What did Peurto Seguro (PS) gain from this contest?",
      "comment_count": 20,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44746,
      "title": "Dataset of public and private LB scores of kernels",
      "comment_count": 7,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44809,
      "title": "Things that didn't work",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44799,
      "title": "Comparing t-SNE implementations",
      "comment_count": 4,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44584,
      "title": "Sharing my pain :-)",
      "comment_count": 13,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44588,
      "title": "Novice questions on blending and stacking",
      "comment_count": 5,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44857,
      "title": "Solution Threads Collection",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44705,
      "title": "submission file sharing",
      "comment_count": 8,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44560,
      "title": "Big fail in two competitions, Lesson learned",
      "comment_count": 11,
      "views": 0,
      "votes": 9
    },
    {
      "id": 44539,
      "title": "well....",
      "comment_count": 20,
      "views": 0,
      "votes": 15
    },
    {
      "id": 44593,
      "title": "The hindsight on cross-validation",
      "comment_count": 8,
      "views": 0,
      "votes": 10
    },
    {
      "id": 44576,
      "title": "When your best model is better than your stack...",
      "comment_count": 9,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40681,
      "title": "Any idea for feature engineering",
      "comment_count": 6,
      "views": 0,
      "votes": 21
    },
    {
      "id": 44531,
      "title": "Best single model performance?",
      "comment_count": 22,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44580,
      "title": "Shake-up",
      "comment_count": 9,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44610,
      "title": "Never trust mix of public kernels",
      "comment_count": 9,
      "views": 0,
      "votes": 3
    },
    {
      "id": 44763,
      "title": "why my rank was deleted???",
      "comment_count": 9,
      "views": 0,
      "votes": null
    },
    {
      "id": 44595,
      "title": "Public kernels sorted by private score?",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 44476,
      "title": "Thank you Kaggle & Porto",
      "comment_count": 7,
      "views": 0,
      "votes": 17
    },
    {
      "id": 44543,
      "title": "The Shakeup - Typical reaction right now",
      "comment_count": 14,
      "views": 0,
      "votes": 4
    },
    {
      "id": 44540,
      "title": "A Newbiee in Kaggle....... a little depressed, a little happy",
      "comment_count": 5,
      "views": 0,
      "votes": 7
    },
    {
      "id": 44551,
      "title": "CV Schemes",
      "comment_count": 5,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44621,
      "title": "39th solution",
      "comment_count": 0,
      "views": 0,
      "votes": 4
    },
    {
      "id": 44532,
      "title": "An Honorable End",
      "comment_count": 9,
      "views": 0,
      "votes": 7
    },
    {
      "id": 44561,
      "title": "LibFFM code",
      "comment_count": 4,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44622,
      "title": "Will be there any change in LB?",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44569,
      "title": "My worst mistake.. (always trust local CV)",
      "comment_count": 3,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44408,
      "title": "Thank you for the biggest competition!",
      "comment_count": 5,
      "views": 0,
      "votes": 13
    },
    {
      "id": 44740,
      "title": "Best single model XGB score on Private LB",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 44605,
      "title": "I'm hooked!",
      "comment_count": 0,
      "views": 0,
      "votes": 7
    },
    {
      "id": 44583,
      "title": "Private score(0.29122). Thank you for kaggler help",
      "comment_count": 1,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44556,
      "title": "50%/50% mixing of home-made/public kernel models worked among my submissions",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44636,
      "title": "TargetEncode in sklearn",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 44554,
      "title": "Suprisingly little overfitting",
      "comment_count": 5,
      "views": 0,
      "votes": 3
    },
    {
      "id": 44461,
      "title": "How many files/folders you have created for this competition?",
      "comment_count": 3,
      "views": 0,
      "votes": 4
    },
    {
      "id": 44447,
      "title": "so... what’s your local CV?",
      "comment_count": 8,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44465,
      "title": "Thank you everyone",
      "comment_count": 2,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44456,
      "title": "Okay to blend the big-gp kernel in the final submission?",
      "comment_count": 17,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44559,
      "title": "what's your could be position",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44136,
      "title": "my neural network got score 0.281!",
      "comment_count": 28,
      "views": 0,
      "votes": 11
    },
    {
      "id": 44604,
      "title": "Public score stayed the same, while CV and private scores were almost equal ",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 44274,
      "title": "Feature Engineering Not Helpful - What am I doing wrong ? ",
      "comment_count": 14,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44653,
      "title": "Baseline for comparing public and private scores.",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43547,
      "title": "Is 0.288 magical and optimal score?",
      "comment_count": 39,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44547,
      "title": "No overfitting",
      "comment_count": 0,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44281,
      "title": "Test set splitting as a \"fork & click\" remedy ?",
      "comment_count": 22,
      "views": 0,
      "votes": 4
    },
    {
      "id": 44609,
      "title": "how is calculated the Private Leaderboard",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 44566,
      "title": "local CV, Public LB and Private LB",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 44565,
      "title": "My 513th place approach",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44563,
      "title": "Local cv works",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 44542,
      "title": "THE SHAKE UP HOLY MOLY",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 44444,
      "title": "Wait 5 hours++ for submission, this is the end!",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 44520,
      "title": "Final submission done, good luck everyone",
      "comment_count": 2,
      "views": 0,
      "votes": 0
    },
    {
      "id": 44200,
      "title": "Issues in Lightgbm",
      "comment_count": 20,
      "views": 0,
      "votes": null
    },
    {
      "id": 44337,
      "title": "Overtraining in Stacking",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44256,
      "title": "Now let's predict shake-up",
      "comment_count": 8,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44480,
      "title": "URL for data Source",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 43760,
      "title": "My guess of the score higher than 287",
      "comment_count": 39,
      "views": 0,
      "votes": 12
    },
    {
      "id": 44474,
      "title": "How to calculate private LB?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 44158,
      "title": "Best single neural network?",
      "comment_count": 13,
      "views": 0,
      "votes": 15
    },
    {
      "id": 43498,
      "title": "Reasons DNN lags behind in this competition",
      "comment_count": 45,
      "views": 0,
      "votes": 5
    },
    {
      "id": 44354,
      "title": "blend weight based on OOF prediction",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44311,
      "title": "How to express the interaction features?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 44126,
      "title": "Happy Thanksgiving!",
      "comment_count": 6,
      "views": 0,
      "votes": 15
    },
    {
      "id": 44441,
      "title": "Brace for Impact !!!",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43741,
      "title": "libFFM modified to Gini-auto-stop with pantiences",
      "comment_count": 23,
      "views": 0,
      "votes": 30
    },
    {
      "id": 44287,
      "title": "random seed for base models vs stacker",
      "comment_count": 9,
      "views": 0,
      "votes": null
    },
    {
      "id": 44295,
      "title": "How to scale predictions when ensemble weights sum more than one?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 42859,
      "title": "Collaborative RAMP team",
      "comment_count": 32,
      "views": 0,
      "votes": 14
    },
    {
      "id": 44267,
      "title": "I used xgboost to run my model but only got LB 0.276",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 43886,
      "title": "Feature discretization (binning)",
      "comment_count": 30,
      "views": 0,
      "votes": 20
    },
    {
      "id": 43467,
      "title": "Why stacking gets good CV but bad LB?",
      "comment_count": 50,
      "views": 0,
      "votes": 9
    },
    {
      "id": 43884,
      "title": "So ... how's everyone doing?",
      "comment_count": 39,
      "views": 0,
      "votes": 3
    },
    {
      "id": 44210,
      "title": "is ‘average kernels’ overfiting LB?",
      "comment_count": 11,
      "views": 0,
      "votes": 1
    },
    {
      "id": 44061,
      "title": "Hard way to get '0.290 on the LB'",
      "comment_count": 10,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44286,
      "title": "What Public LB scores are realistically \"in the running\"?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43779,
      "title": "4th Digit After the Decimal Point?",
      "comment_count": 22,
      "views": 0,
      "votes": 22
    },
    {
      "id": 44123,
      "title": "Who is gonna break it???",
      "comment_count": 2,
      "views": 0,
      "votes": 6
    },
    {
      "id": 43933,
      "title": "It's time to play the LB shake-up prediction game again!",
      "comment_count": 4,
      "views": 0,
      "votes": 6
    },
    {
      "id": 44194,
      "title": "What are your loss function of linear model for ensembling?  ",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 44005,
      "title": "2nd most competitors/teams in any competition?",
      "comment_count": 8,
      "views": 0,
      "votes": 2
    },
    {
      "id": 44167,
      "title": "after the competition ends...",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43357,
      "title": "Decision time",
      "comment_count": 13,
      "views": 0,
      "votes": 31
    },
    {
      "id": 44093,
      "title": "is it possible for single model to get above 0.286LB?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 44011,
      "title": "Whats the role of seed in model training?",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43798,
      "title": "My Public/Private Scores [Santander]",
      "comment_count": 9,
      "views": 0,
      "votes": 8
    },
    {
      "id": 44020,
      "title": "Intuitive and fast to compute gini",
      "comment_count": 2,
      "views": 0,
      "votes": 5
    },
    {
      "id": 43453,
      "title": "Ho (dis)similar are train and test data?",
      "comment_count": 18,
      "views": 0,
      "votes": 70
    },
    {
      "id": 44089,
      "title": "More than 5 submissions today?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43891,
      "title": "rank rises like a rocket",
      "comment_count": 10,
      "views": 0,
      "votes": null
    },
    {
      "id": 44027,
      "title": "Looking for last minute teammate with something original",
      "comment_count": 11,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43745,
      "title": "How to save and load a TensorFlow model",
      "comment_count": 7,
      "views": 0,
      "votes": null
    },
    {
      "id": 43140,
      "title": "A Kaggler's Guide to Model Stacking in Practice",
      "comment_count": 19,
      "views": 0,
      "votes": 25
    },
    {
      "id": 44014,
      "title": "Why team merge failure?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42804,
      "title": "5 things I learned from this competition",
      "comment_count": 18,
      "views": 0,
      "votes": 77
    },
    {
      "id": 43808,
      "title": "a random forest problem confused me",
      "comment_count": 10,
      "views": 0,
      "votes": null
    },
    {
      "id": 43797,
      "title": "Which CV score to trust if it is varying every time?",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43975,
      "title": "To Inflate or deflate features?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 43297,
      "title": "why I'm not trying high rank public kernels",
      "comment_count": 8,
      "views": 0,
      "votes": 11
    },
    {
      "id": 43447,
      "title": "My Approach to Kaggle Competitions",
      "comment_count": 20,
      "views": 0,
      "votes": 17
    },
    {
      "id": 43813,
      "title": "The meanings of column names",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43850,
      "title": "Oversampling/undersampling",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42823,
      "title": "Is there a reason why people aren't publishing kernels for LB>=0.285?",
      "comment_count": 29,
      "views": 0,
      "votes": 11
    },
    {
      "id": 43481,
      "title": "For those high rank competitors",
      "comment_count": 19,
      "views": 0,
      "votes": 5
    },
    {
      "id": 43288,
      "title": "LB 0.291",
      "comment_count": 21,
      "views": 0,
      "votes": 41
    },
    {
      "id": 43075,
      "title": "The insurance company needs the predicted probabilities not the predicted rankings",
      "comment_count": 33,
      "views": 0,
      "votes": 6
    },
    {
      "id": 43336,
      "title": "Additional priza - \"Overfitting expert\"",
      "comment_count": 9,
      "views": 0,
      "votes": 22
    },
    {
      "id": 43384,
      "title": "Is 0.001 a noise level?",
      "comment_count": 23,
      "views": 0,
      "votes": 5
    },
    {
      "id": 43775,
      "title": "Difference in predicted probability ",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 43766,
      "title": "Team Up!!",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 43657,
      "title": "Why the most people don't use xgboost.cv(...) for CV?",
      "comment_count": 5,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43373,
      "title": "New style kernels page",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43731,
      "title": "quite remarkable variance of gini",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 43647,
      "title": "Why would anyone use a ready-made public kernel?",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 43364,
      "title": "A short story about last weekend",
      "comment_count": 4,
      "views": 0,
      "votes": 15
    },
    {
      "id": 42197,
      "title": "Why are we struggling? An illustrated exploration ...",
      "comment_count": 35,
      "views": 0,
      "votes": 129
    },
    {
      "id": 43315,
      "title": "Big Shake Up on Private LB!",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 43001,
      "title": "For newbies",
      "comment_count": 14,
      "views": 0,
      "votes": 11
    },
    {
      "id": 43410,
      "title": "Current LB distribution",
      "comment_count": 11,
      "views": 0,
      "votes": 6
    },
    {
      "id": 43653,
      "title": "Some questions about target imbalance and feature engineering",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 43608,
      "title": "what  feature engineering methods are applicable on this dataset! ",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 43633,
      "title": "do you  want to come with me on LB",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43455,
      "title": "2 weeks to go and I am still in the woods",
      "comment_count": 8,
      "views": 0,
      "votes": 3
    },
    {
      "id": 43720,
      "title": "Columns are so Noisy!",
      "comment_count": 0,
      "views": 0,
      "votes": -17
    },
    {
      "id": 43282,
      "title": "Tensorflow problem: disconnect between accuracy and gini ",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 43495,
      "title": "Gini standard error as a function of data size",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43492,
      "title": "how to determine which variables are noise",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43483,
      "title": "How to make our models spit out probabilities?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 40599,
      "title": "ps_car_15 are square root of integers",
      "comment_count": 77,
      "views": 0,
      "votes": 68
    },
    {
      "id": 43339,
      "title": "How do you guys run rgf classifier?",
      "comment_count": 7,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43191,
      "title": "Sharing a very intuitive explanation to gini coefficient!",
      "comment_count": 2,
      "views": 0,
      "votes": 4
    },
    {
      "id": 42982,
      "title": "Target value not in test data",
      "comment_count": 6,
      "views": 0,
      "votes": -6
    },
    {
      "id": 43194,
      "title": "Looking for teammate(s)",
      "comment_count": 4,
      "views": 0,
      "votes": 5
    },
    {
      "id": 43439,
      "title": "Hyperparameter Tuning for Tensorflow",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 43383,
      "title": "How to limit the number of threads used for Keras?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 43429,
      "title": " How can we become confident that our stacking or ensembling technique is not over-fitting Public LB? ",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43314,
      "title": "How can get the form asked?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 43323,
      "title": "Why do you do your best for Kaggle?",
      "comment_count": 5,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43344,
      "title": "Strategy to select the final submissions",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43376,
      "title": "Something, something, danger zone",
      "comment_count": 1,
      "views": 0,
      "votes": 2
    },
    {
      "id": 43294,
      "title": "actively tuning but no submisison yet",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 42785,
      "title": "How to stabilize score variance?",
      "comment_count": 24,
      "views": 0,
      "votes": 5
    },
    {
      "id": 43451,
      "title": "Data transformation?",
      "comment_count": 0,
      "views": 0,
      "votes": -2
    },
    {
      "id": 43073,
      "title": "how to weight your models",
      "comment_count": 15,
      "views": 0,
      "votes": 6
    },
    {
      "id": 43393,
      "title": "How to avoid overfitting to a specific random seed?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42816,
      "title": "Model selection",
      "comment_count": 11,
      "views": 0,
      "votes": 15
    },
    {
      "id": 43488,
      "title": "hmmm, does this competition counted medal?",
      "comment_count": 0,
      "views": 0,
      "votes": -6
    },
    {
      "id": 42954,
      "title": "How to Know Who You Can Trust in This Competition",
      "comment_count": 8,
      "views": 0,
      "votes": 19
    },
    {
      "id": 43248,
      "title": "Why is calculated Gini not reflected in Kaggle score?",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43398,
      "title": "Why is ranks so effective?",
      "comment_count": 0,
      "views": 0,
      "votes": -2
    },
    {
      "id": 43308,
      "title": "Any one in Top 100 Looking for Team Mates?",
      "comment_count": 2,
      "views": 0,
      "votes": -1
    },
    {
      "id": 43322,
      "title": "The LB algorithm",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 43304,
      "title": "StratifiedKFold problem?",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42853,
      "title": "How to select best n_rounds for prediction (boosting models)",
      "comment_count": 25,
      "views": 0,
      "votes": 10
    },
    {
      "id": 42735,
      "title": "Not a classification problem",
      "comment_count": 43,
      "views": 0,
      "votes": 25
    },
    {
      "id": 41037,
      "title": "Cross-validation issues",
      "comment_count": 81,
      "views": 0,
      "votes": 38
    },
    {
      "id": 43067,
      "title": "Building xgboost with GPU support",
      "comment_count": 8,
      "views": 0,
      "votes": 6
    },
    {
      "id": 43205,
      "title": "Score 0 on public LB",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42957,
      "title": "Seeking teammate(s) with something interesting",
      "comment_count": 7,
      "views": 0,
      "votes": 7
    },
    {
      "id": 42861,
      "title": "Dealing with ordinal variables",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41165,
      "title": "Number of threads impact on the score",
      "comment_count": 12,
      "views": 0,
      "votes": null
    },
    {
      "id": 42709,
      "title": "FTLR - 0.258 - Good for Ensemble",
      "comment_count": 50,
      "views": 0,
      "votes": 11
    },
    {
      "id": 42526,
      "title": "Happy Halloween! Here is something really scary.",
      "comment_count": 9,
      "views": 0,
      "votes": 31
    },
    {
      "id": 41504,
      "title": "How to do with different answers?",
      "comment_count": 2,
      "views": 0,
      "votes": -8
    },
    {
      "id": 42675,
      "title": "New Genetic algorithm I've been working on (pictures)",
      "comment_count": 8,
      "views": 0,
      "votes": 2
    },
    {
      "id": 42841,
      "title": "How to improve after reaching 0.287?",
      "comment_count": 50,
      "views": 0,
      "votes": -9
    },
    {
      "id": 42693,
      "title": "What's the best rate of ensemble?",
      "comment_count": 14,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40728,
      "title": "ps_ind_06-10 'binary' variables are one-hot encoded",
      "comment_count": 16,
      "views": 0,
      "votes": 32
    },
    {
      "id": 42327,
      "title": "Interesting Pattern (LB)",
      "comment_count": 37,
      "views": 0,
      "votes": 5
    },
    {
      "id": 42995,
      "title": "Please share any neural net implementation for classification",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 43175,
      "title": "dgagsag",
      "comment_count": 1,
      "views": 0,
      "votes": -11
    },
    {
      "id": 42556,
      "title": "Cross-validation standard deviation",
      "comment_count": 4,
      "views": 0,
      "votes": 21
    },
    {
      "id": 41238,
      "title": "Competition workflow",
      "comment_count": 25,
      "views": 0,
      "votes": 10
    },
    {
      "id": 42746,
      "title": "After Stacking...what next?",
      "comment_count": 6,
      "views": 0,
      "votes": 4
    },
    {
      "id": 43110,
      "title": "TEST",
      "comment_count": 0,
      "views": 0,
      "votes": -5
    },
    {
      "id": 42920,
      "title": "bagging problem",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 42457,
      "title": "Any tips to reach a LB score >  0.286 :( ? ",
      "comment_count": 42,
      "views": 0,
      "votes": 6
    },
    {
      "id": 42977,
      "title": "Is heplful for you to address imbalance of dataset?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42776,
      "title": "on the meaning of \"overfit\"",
      "comment_count": 6,
      "views": 0,
      "votes": 2
    },
    {
      "id": 42862,
      "title": "Dealing with missing values",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 43054,
      "title": "Bagging",
      "comment_count": 0,
      "views": 0,
      "votes": -5
    },
    {
      "id": 42950,
      "title": "Feature Meanings",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40540,
      "title": "Single model performance",
      "comment_count": 50,
      "views": 0,
      "votes": 15
    },
    {
      "id": 42924,
      "title": "Passing ANN predictions as a feature into LightGBM/XGBoost?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42522,
      "title": "xgboost CPU usage",
      "comment_count": 12,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41562,
      "title": "Kaggle Tutorial on this competition",
      "comment_count": 23,
      "views": 0,
      "votes": 44
    },
    {
      "id": 42860,
      "title": "model validation",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42851,
      "title": "ParserError",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42545,
      "title": "Gini with Keras",
      "comment_count": 8,
      "views": 0,
      "votes": null
    },
    {
      "id": 42822,
      "title": "XGboost Predict",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 42811,
      "title": "Is there a detailed description of the variables in the data sets?",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42582,
      "title": "whether to focus on stacking or better single models?",
      "comment_count": 3,
      "views": 0,
      "votes": 7
    },
    {
      "id": 42739,
      "title": "issue with different methods of gini score calculates",
      "comment_count": 1,
      "views": 0,
      "votes": 4
    },
    {
      "id": 42700,
      "title": "Is features already engineered?",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 42787,
      "title": "Under what conditions/algorithms, the data would overfit?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42581,
      "title": "ps_car_13 - restored to integers",
      "comment_count": 8,
      "views": 0,
      "votes": 5
    },
    {
      "id": 42691,
      "title": "New features, XGB upsampling + BayesOpt",
      "comment_count": 0,
      "views": 0,
      "votes": 7
    },
    {
      "id": 42771,
      "title": "Error to submission",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 42770,
      "title": "Determining optimal weights in model averaging",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42186,
      "title": "How to stack",
      "comment_count": 17,
      "views": 0,
      "votes": 11
    },
    {
      "id": 42553,
      "title": "average of gini scores",
      "comment_count": 6,
      "views": 0,
      "votes": 6
    },
    {
      "id": 42726,
      "title": "how many members can be on a team ",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42747,
      "title": "How to write kernels?",
      "comment_count": 2,
      "views": 0,
      "votes": -1
    },
    {
      "id": 42416,
      "title": "How do you organize your code?",
      "comment_count": 7,
      "views": 0,
      "votes": 5
    },
    {
      "id": 42734,
      "title": "Ensemble by Aggregating Rank",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42671,
      "title": "Variable are not matched",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42603,
      "title": "There is no need to normalize gini",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 40969,
      "title": "Why the best results get stuck around 0.28?",
      "comment_count": 8,
      "views": 0,
      "votes": 5
    },
    {
      "id": 42328,
      "title": "Is regularized hyperparameter optimization a thing?",
      "comment_count": 10,
      "views": 0,
      "votes": 6
    },
    {
      "id": 40424,
      "title": "[Beginner] Want to join a team.",
      "comment_count": 22,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42625,
      "title": "Why the features are anonymized?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 42360,
      "title": "Imbalanced data set",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 42291,
      "title": "How to stack to get better score(above 0.284)?",
      "comment_count": 8,
      "views": 0,
      "votes": 4
    },
    {
      "id": 40760,
      "title": "How did people with .283 and above manage to get a high score ",
      "comment_count": 58,
      "views": 0,
      "votes": 30
    },
    {
      "id": 42550,
      "title": "about Feature Importance ",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41564,
      "title": "What to include in the submission file ??",
      "comment_count": 20,
      "views": 0,
      "votes": null
    },
    {
      "id": 42535,
      "title": "ps_car_06_cat could be the rate category",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42315,
      "title": "ps_car_11_cat - week of year?",
      "comment_count": 16,
      "views": 0,
      "votes": null
    },
    {
      "id": 42443,
      "title": "What wrong with this category encode?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 42240,
      "title": "Agnostic Feature Engineering - Package ?",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 42300,
      "title": "Can representation learning help to improve xgboost?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 42085,
      "title": "Breaking out of 0.282",
      "comment_count": 13,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41844,
      "title": "Great work ireallyloveyou",
      "comment_count": 14,
      "views": 0,
      "votes": 21
    },
    {
      "id": 42461,
      "title": "Loss function",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 42312,
      "title": "Best practices of Feature Engineering",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41716,
      "title": "Should we replace -1 with other val?",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 41658,
      "title": "Try harmonic Mean!",
      "comment_count": 23,
      "views": 0,
      "votes": 14
    },
    {
      "id": 42338,
      "title": "How high single model performance can we get?",
      "comment_count": 5,
      "views": 0,
      "votes": -4
    },
    {
      "id": 41488,
      "title": "ps_car_12 - vehicle engine cylinder capacity (cc)",
      "comment_count": 9,
      "views": 0,
      "votes": 35
    },
    {
      "id": 41907,
      "title": "ps_ind_14 is sum of 4 features",
      "comment_count": 10,
      "views": 0,
      "votes": 19
    },
    {
      "id": 42201,
      "title": "\"age\" feature",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 42181,
      "title": "Regarding improvement in the model",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 42079,
      "title": "logloss and auc both strangely decreasing!",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 41445,
      "title": "best single model( bagging included)",
      "comment_count": 32,
      "views": 0,
      "votes": 17
    },
    {
      "id": 41784,
      "title": "Any tips for which models should be used in stacking?",
      "comment_count": 9,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41908,
      "title": "Scored .241 by simply using plain random forest",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41970,
      "title": "Why removing ps_calc features improve the results?",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 41595,
      "title": "Should we eliminated any features?",
      "comment_count": 24,
      "views": 0,
      "votes": 6
    },
    {
      "id": 41556,
      "title": "No effect on feature engineering",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 41534,
      "title": "New Keras script (LB 0.270 after modifications)",
      "comment_count": 17,
      "views": 0,
      "votes": 8
    },
    {
      "id": 42086,
      "title": "nnet old  school 0.275",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41897,
      "title": "Cannot upload a prediction",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 40545,
      "title": "Gini score Nan in xgboost",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 41487,
      "title": "Keep or discard - confusing decision",
      "comment_count": 19,
      "views": 0,
      "votes": 6
    },
    {
      "id": 41617,
      "title": "Can Normalized Gini coefficient be negative?",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41633,
      "title": "Fully connected NN instead of RF helped to increase accuracy up to 100%",
      "comment_count": 1,
      "views": 0,
      "votes": 4
    },
    {
      "id": 41656,
      "title": "Why one-hot encoding does not enhance my xgboost?",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 40871,
      "title": "ps_car_11 and ps_car11_cat",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41643,
      "title": "Could it be some information leakeage???",
      "comment_count": 8,
      "views": 0,
      "votes": null
    },
    {
      "id": 40409,
      "title": "CV vs. LB",
      "comment_count": 20,
      "views": 0,
      "votes": 10
    },
    {
      "id": 41695,
      "title": "a day in the life...",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41693,
      "title": "Anyone using MCMC here?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 40942,
      "title": "Is there ang lib of FM(python) for windows?",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 41463,
      "title": "Team Up",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41591,
      "title": "h2o Help",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41594,
      "title": "Does \"not an improvement of your best score\" include extra digits?",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 41440,
      "title": "How to Treat Missing Values",
      "comment_count": 4,
      "views": 0,
      "votes": 4
    },
    {
      "id": 41748,
      "title": "Good work",
      "comment_count": 0,
      "views": 0,
      "votes": -16
    },
    {
      "id": 41302,
      "title": "Numerical order similarity : absolute value of correlation coefficient vs feature importance",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41517,
      "title": "Curious behaviour of the Gini Normal Coefficient",
      "comment_count": 4,
      "views": 0,
      "votes": 2
    },
    {
      "id": 41518,
      "title": "Stacknet ?? ",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41548,
      "title": "Stucked at 0.26",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41200,
      "title": "ps_reg_03",
      "comment_count": 11,
      "views": 0,
      "votes": 13
    },
    {
      "id": 41612,
      "title": "parameters defining the missing value",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41663,
      "title": "anyone added this feature(ps_car_14*ps_ind_15) to the model? ",
      "comment_count": 0,
      "views": 0,
      "votes": -7
    },
    {
      "id": 41161,
      "title": "Feature Engineering using correlation",
      "comment_count": 14,
      "views": 0,
      "votes": 2
    },
    {
      "id": 41536,
      "title": "Classification results in only 1s and 0s",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41395,
      "title": "Calculating the normalized gini coefficient",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 41379,
      "title": "external data allowed?",
      "comment_count": 3,
      "views": 0,
      "votes": 5
    },
    {
      "id": 41454,
      "title": "Comparing drivers",
      "comment_count": 8,
      "views": 0,
      "votes": null
    },
    {
      "id": 41394,
      "title": "Stacking: individual model accuracy",
      "comment_count": 8,
      "views": 0,
      "votes": 2
    },
    {
      "id": 41362,
      "title": "Does t-SNE help?",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41399,
      "title": "Female Demographics",
      "comment_count": 9,
      "views": 0,
      "votes": null
    },
    {
      "id": 41269,
      "title": "How to improve score from 0.281",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41311,
      "title": "Validation Set",
      "comment_count": 8,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41288,
      "title": "Level 2 Classifier Issues",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41433,
      "title": "Train/Test Disparity",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40463,
      "title": "How are you predicting the class after an year ?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41057,
      "title": "Car insurance quote from Porto Seguro",
      "comment_count": 0,
      "views": 0,
      "votes": 28
    },
    {
      "id": 41393,
      "title": "Should we address the imbalance of trainData?",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40830,
      "title": "Guessing the missing values",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 41400,
      "title": "Do not classify but submit probability values",
      "comment_count": 0,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41237,
      "title": "Rounding of scores in the leaderboard",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41066,
      "title": "How do you investigate/estimate models that deal with so much data?",
      "comment_count": 5,
      "views": 0,
      "votes": 10
    },
    {
      "id": 41108,
      "title": "AUC calculation in TensorFlow (keras)",
      "comment_count": 5,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41015,
      "title": "Using Gini or auc with keras",
      "comment_count": 5,
      "views": 0,
      "votes": 2
    },
    {
      "id": 41136,
      "title": "Need Stacking starter script without using sklearn ..",
      "comment_count": 19,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41354,
      "title": "Andre  Ng L Neural Net From Coursera Specialization",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41296,
      "title": "Data Explanations",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40748,
      "title": "Features already engineered",
      "comment_count": 21,
      "views": 0,
      "votes": 31
    },
    {
      "id": 41329,
      "title": "Always got 2.90+ score on validation set(25% of the training data)...",
      "comment_count": 3,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40706,
      "title": "On Gini metrics",
      "comment_count": 19,
      "views": 0,
      "votes": 14
    },
    {
      "id": 40800,
      "title": "Rescore?",
      "comment_count": 33,
      "views": 0,
      "votes": 4
    },
    {
      "id": 41300,
      "title": "simplest way to downsample pandas dataframe based on outcome variable",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41164,
      "title": "PyTorch Deep Learning starter: 0.258",
      "comment_count": 2,
      "views": 0,
      "votes": 7
    },
    {
      "id": 41196,
      "title": "how to have a perfect random_state ？",
      "comment_count": 6,
      "views": 0,
      "votes": -3
    },
    {
      "id": 41320,
      "title": "New  script -Stacked Features  with Shallow Neural Net",
      "comment_count": 2,
      "views": 0,
      "votes": -3
    },
    {
      "id": 40826,
      "title": "Problem in submission",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 40509,
      "title": "Normalized Gini Coefficient",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 41274,
      "title": "Why is the target column in test data missing?",
      "comment_count": 2,
      "views": 0,
      "votes": -5
    },
    {
      "id": 40763,
      "title": "how to implement gini loss for keras?",
      "comment_count": 12,
      "views": 0,
      "votes": null
    },
    {
      "id": 40797,
      "title": "Problems with LightGBM?",
      "comment_count": 8,
      "views": 0,
      "votes": 4
    },
    {
      "id": 41194,
      "title": "Kinetic Aggregated with forza  baseline for diversity",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41187,
      "title": "Looking for a Python Coding Guy to work as  a Team on this Safe Driver Prediction Challenge",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41125,
      "title": "Don't use Xgboost-GPU",
      "comment_count": 3,
      "views": 0,
      "votes": null
    },
    {
      "id": 40774,
      "title": "External Papers",
      "comment_count": 1,
      "views": 0,
      "votes": 23
    },
    {
      "id": 40734,
      "title": "Feature Engineering with unknown data",
      "comment_count": 17,
      "views": 0,
      "votes": 10
    },
    {
      "id": 41140,
      "title": "Class Imbalance",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41097,
      "title": "How to deal with inbalanced data",
      "comment_count": 1,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40950,
      "title": "XGBoost parameters tuning",
      "comment_count": 6,
      "views": 0,
      "votes": 3
    },
    {
      "id": 41072,
      "title": "Differences in CV fold scores",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 41032,
      "title": "Reducing training time",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 41033,
      "title": "data set confusion",
      "comment_count": 1,
      "views": 0,
      "votes": -12
    },
    {
      "id": 41175,
      "title": "Please confirm/disconfirm interpretation of variables",
      "comment_count": 2,
      "views": 0,
      "votes": -7
    },
    {
      "id": 40392,
      "title": "Decimals",
      "comment_count": 10,
      "views": 0,
      "votes": 30
    },
    {
      "id": 41045,
      "title": "Predicted target values are all zero",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40765,
      "title": "Data Distribution and its nature.",
      "comment_count": 9,
      "views": 0,
      "votes": 5
    },
    {
      "id": 40881,
      "title": "What is cross validation, and how to use it?",
      "comment_count": 4,
      "views": 0,
      "votes": null
    },
    {
      "id": 41020,
      "title": "Reg columns with too many missing values",
      "comment_count": 3,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41010,
      "title": "Simple Ways to Improve LB Accuracy.",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 40990,
      "title": "Problem about hardware",
      "comment_count": 6,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40835,
      "title": "xgboost prediction in R is different on sparse and normal matrices",
      "comment_count": 1,
      "views": 0,
      "votes": 5
    },
    {
      "id": 40978,
      "title": "Association rule learning / Apriori algorithm",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 41031,
      "title": "data set uploading",
      "comment_count": 0,
      "views": 0,
      "votes": -3
    },
    {
      "id": 40756,
      "title": "Searching  hard working  partner",
      "comment_count": 11,
      "views": 0,
      "votes": null
    },
    {
      "id": 40840,
      "title": "Normalization of features",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40874,
      "title": "Feature Engineering",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40576,
      "title": "Alternative Gini calculation function",
      "comment_count": 3,
      "views": 0,
      "votes": 17
    },
    {
      "id": 40794,
      "title": "Even faster gini computation",
      "comment_count": 0,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40474,
      "title": "Data Dictionary?",
      "comment_count": 3,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40610,
      "title": "Looking for someone to form a team.",
      "comment_count": 3,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40551,
      "title": "-1 in float columns",
      "comment_count": 9,
      "views": 0,
      "votes": 10
    },
    {
      "id": 40501,
      "title": " Any clue  to  break   the 0.281  barrier?",
      "comment_count": 17,
      "views": 0,
      "votes": 2
    },
    {
      "id": 40684,
      "title": "Are ps_car_12, ps_car_13, and ps_car_14 logit transforamtions?",
      "comment_count": 1,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40682,
      "title": "Is there any statistical decision theory for optimizing Gini?",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40617,
      "title": "Classification vs Regression",
      "comment_count": 2,
      "views": 0,
      "votes": 3
    },
    {
      "id": 40785,
      "title": "Kernel  0.482 ,  run  and  go up the board",
      "comment_count": 3,
      "views": 0,
      "votes": -13
    },
    {
      "id": 40620,
      "title": "Since I have a best model,why should I ensemble other models?",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40631,
      "title": "Chose stacking but it performs badly..How to chose base models?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40638,
      "title": "how to get xgboost eval as simple number?",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 40596,
      "title": "Data Collection",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40583,
      "title": "High Risk Driver* looking for a teammate!",
      "comment_count": 6,
      "views": 0,
      "votes": null
    },
    {
      "id": 40338,
      "title": "Friendly partner  with top  score",
      "comment_count": 5,
      "views": 0,
      "votes": null
    },
    {
      "id": 40568,
      "title": "Data Privacy",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40459,
      "title": "Something wrong with the competition metrics?",
      "comment_count": 9,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40531,
      "title": "Submission Error",
      "comment_count": 2,
      "views": 0,
      "votes": null
    },
    {
      "id": 40598,
      "title": "The ID column for prediction",
      "comment_count": 2,
      "views": 0,
      "votes": -4
    },
    {
      "id": 40425,
      "title": "Anyone facing issue with data download links ?",
      "comment_count": 5,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40410,
      "title": "Brazil has 7 levels of bonus-malus",
      "comment_count": 0,
      "views": 0,
      "votes": 5
    },
    {
      "id": 40510,
      "title": "What is the equation to calculate Normalized Gini Coefficient?",
      "comment_count": 1,
      "views": 0,
      "votes": null
    },
    {
      "id": 40440,
      "title": "is stuff in Kernels free for me to copy and used in my work?",
      "comment_count": 4,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40491,
      "title": "What About The Data Quality?",
      "comment_count": 2,
      "views": 0,
      "votes": -3
    },
    {
      "id": 40469,
      "title": "Sample Submission",
      "comment_count": 2,
      "views": 0,
      "votes": 1
    },
    {
      "id": 40416,
      "title": "ps_car_01_cat is in test but not in train",
      "comment_count": 0,
      "views": 0,
      "votes": 0
    },
    {
      "id": 40562,
      "title": "how to load images for prediction in tensorflow",
      "comment_count": 1,
      "views": 0,
      "votes": -18
    }
  ],
  "errors": []
}