You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@hivemall.apache.org by my...@apache.org on 2017/06/15 10:36:06 UTC
[21/51] [partial] incubator-hivemall-site git commit: Updated
userguide for dimsum and general classifier/regressor
http://git-wip-us.apache.org/repos/asf/incubator-hivemall-site/blob/f103c424/userguide/misc/prediction.html
----------------------------------------------------------------------
diff --git a/userguide/misc/prediction.html b/userguide/misc/prediction.html
new file mode 100644
index 0000000..60204e5
--- /dev/null
+++ b/userguide/misc/prediction.html
@@ -0,0 +1,2370 @@
+
+<!DOCTYPE HTML>
+<html lang="" >
+ <head>
+ <meta charset="UTF-8">
+ <meta content="text/html; charset=utf-8" http-equiv="Content-Type">
+ <title>How Prediction Works ยท Hivemall User Manual</title>
+ <meta http-equiv="X-UA-Compatible" content="IE=edge" />
+ <meta name="description" content="">
+ <meta name="generator" content="GitBook 3.2.2">
+
+
+
+
+ <link rel="stylesheet" href="../gitbook/style.css">
+
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-splitter/splitter.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-etoc/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-callouts/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-toggle-chapters/toggle.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-codeblock-filename/block.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-expandable-chapters/expandable-chapters.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-multipart/multipart.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-katex/katex.min.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-emphasize/plugin.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-highlight/website.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-search/search.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-fontsettings/website.css">
+
+
+
+ <link rel="stylesheet" href="../gitbook/gitbook-plugin-theme-api/theme-api.css">
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ <meta name="HandheldFriendly" content="true"/>
+ <meta name="viewport" content="width=device-width, initial-scale=1, user-scalable=no">
+ <meta name="apple-mobile-web-app-capable" content="yes">
+ <meta name="apple-mobile-web-app-status-bar-style" content="black">
+ <link rel="apple-touch-icon-precomposed" sizes="152x152" href="../gitbook/images/apple-touch-icon-precomposed-152.png">
+ <link rel="shortcut icon" href="../gitbook/images/favicon.ico" type="image/x-icon">
+
+
+ <link rel="next" href="../regression/general.html" />
+
+
+ <link rel="prev" href="../eval/lr_datagen.html" />
+
+
+ </head>
+ <body>
+
+<div class="book">
+ <div class="book-summary">
+
+
+<div id="book-search-input" role="search">
+ <input type="text" placeholder="Type to search" />
+</div>
+
+
+ <nav role="navigation">
+
+
+
+<ul class="summary">
+
+
+
+
+ <li>
+ <a href="http://hivemall.incubator.apache.org/" target="_blank" class="custom-link"><i class="fa fa-home"></i> Home</a>
+ </li>
+
+
+
+
+ <li class="divider"></li>
+
+
+
+
+ <li class="header">TABLE OF CONTENTS</li>
+
+
+
+ <li class="chapter " data-level="1.1" data-path="../">
+
+ <a href="../">
+
+
+ <b>1.1.</b>
+
+ Introduction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2" data-path="../getting_started/">
+
+ <a href="../getting_started/">
+
+
+ <b>1.2.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.2.1" data-path="../getting_started/installation.html">
+
+ <a href="../getting_started/installation.html">
+
+
+ <b>1.2.1.</b>
+
+ Installation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2.2" data-path="../getting_started/permanent-functions.html">
+
+ <a href="../getting_started/permanent-functions.html">
+
+
+ <b>1.2.2.</b>
+
+ Install as permanent functions
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.2.3" data-path="../getting_started/input-format.html">
+
+ <a href="../getting_started/input-format.html">
+
+
+ <b>1.2.3.</b>
+
+ Input Format
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.3" data-path="../tips/">
+
+ <a href="../tips/">
+
+
+ <b>1.3.</b>
+
+ Tips for Effective Hivemall
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.3.1" data-path="../tips/addbias.html">
+
+ <a href="../tips/addbias.html">
+
+
+ <b>1.3.1.</b>
+
+ Explicit addBias() for better prediction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.2" data-path="../tips/rand_amplify.html">
+
+ <a href="../tips/rand_amplify.html">
+
+
+ <b>1.3.2.</b>
+
+ Use rand_amplify() to better prediction results
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.3" data-path="../tips/rt_prediction.html">
+
+ <a href="../tips/rt_prediction.html">
+
+
+ <b>1.3.3.</b>
+
+ Real-time Prediction on RDBMS
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.4" data-path="../tips/ensemble_learning.html">
+
+ <a href="../tips/ensemble_learning.html">
+
+
+ <b>1.3.4.</b>
+
+ Ensemble learning for stable prediction
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.5" data-path="../tips/mixserver.html">
+
+ <a href="../tips/mixserver.html">
+
+
+ <b>1.3.5.</b>
+
+ Mixing models for a better prediction convergence (MIX server)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.3.6" data-path="../tips/emr.html">
+
+ <a href="../tips/emr.html">
+
+
+ <b>1.3.6.</b>
+
+ Run Hivemall on Amazon Elastic MapReduce
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.4" data-path="../tips/general_tips.html">
+
+ <a href="../tips/general_tips.html">
+
+
+ <b>1.4.</b>
+
+ General Hive/Hadoop tips
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.4.1" data-path="../tips/rowid.html">
+
+ <a href="../tips/rowid.html">
+
+
+ <b>1.4.1.</b>
+
+ Adding rowid for each row
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.4.2" data-path="../tips/hadoop_tuning.html">
+
+ <a href="../tips/hadoop_tuning.html">
+
+
+ <b>1.4.2.</b>
+
+ Hadoop tuning for Hivemall
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="1.5" data-path="../troubleshooting/">
+
+ <a href="../troubleshooting/">
+
+
+ <b>1.5.</b>
+
+ Troubleshooting
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="1.5.1" data-path="../troubleshooting/oom.html">
+
+ <a href="../troubleshooting/oom.html">
+
+
+ <b>1.5.1.</b>
+
+ OutOfMemoryError in training
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.2" data-path="../troubleshooting/mapjoin_task_error.html">
+
+ <a href="../troubleshooting/mapjoin_task_error.html">
+
+
+ <b>1.5.2.</b>
+
+ SemanticException Generate Map Join Task Error: Cannot serialize object
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.3" data-path="../troubleshooting/asterisk.html">
+
+ <a href="../troubleshooting/asterisk.html">
+
+
+ <b>1.5.3.</b>
+
+ Asterisk argument for UDTF does not work
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.4" data-path="../troubleshooting/num_mappers.html">
+
+ <a href="../troubleshooting/num_mappers.html">
+
+
+ <b>1.5.4.</b>
+
+ The number of mappers is less than input splits in Hadoop 2.x
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="1.5.5" data-path="../troubleshooting/mapjoin_classcastex.html">
+
+ <a href="../troubleshooting/mapjoin_classcastex.html">
+
+
+ <b>1.5.5.</b>
+
+ Map-side Join causes ClassCastException on Tez
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part II - Generic Features</li>
+
+
+
+ <li class="chapter " data-level="2.1" data-path="generic_funcs.html">
+
+ <a href="generic_funcs.html">
+
+
+ <b>2.1.</b>
+
+ List of generic Hivemall functions
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="2.2" data-path="topk.html">
+
+ <a href="topk.html">
+
+
+ <b>2.2.</b>
+
+ Efficient Top-K query processing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="2.3" data-path="tokenizer.html">
+
+ <a href="tokenizer.html">
+
+
+ <b>2.3.</b>
+
+ English/Japanese Text Tokenizer
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part III - Feature Engineering</li>
+
+
+
+ <li class="chapter " data-level="3.1" data-path="../ft_engineering/scaling.html">
+
+ <a href="../ft_engineering/scaling.html">
+
+
+ <b>3.1.</b>
+
+ Feature Scaling
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.2" data-path="../ft_engineering/hashing.html">
+
+ <a href="../ft_engineering/hashing.html">
+
+
+ <b>3.2.</b>
+
+ Feature Hashing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.3" data-path="../ft_engineering/selection.html">
+
+ <a href="../ft_engineering/selection.html">
+
+
+ <b>3.3.</b>
+
+ Feature Selection
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.4" data-path="../ft_engineering/binning.html">
+
+ <a href="../ft_engineering/binning.html">
+
+
+ <b>3.4.</b>
+
+ Feature Binning
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.5" data-path="../ft_engineering/tfidf.html">
+
+ <a href="../ft_engineering/tfidf.html">
+
+
+ <b>3.5.</b>
+
+ TF-IDF Calculation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.6" data-path="../ft_engineering/ft_trans.html">
+
+ <a href="../ft_engineering/ft_trans.html">
+
+
+ <b>3.6.</b>
+
+ FEATURE TRANSFORMATION
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="3.6.1" data-path="../ft_engineering/vectorization.html">
+
+ <a href="../ft_engineering/vectorization.html">
+
+
+ <b>3.6.1.</b>
+
+ Feature Vectorization
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="3.6.2" data-path="../ft_engineering/quantify.html">
+
+ <a href="../ft_engineering/quantify.html">
+
+
+ <b>3.6.2.</b>
+
+ Quantify non-number features
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part IV - Evaluation</li>
+
+
+
+ <li class="chapter " data-level="4.1" data-path="../eval/stat_eval.html">
+
+ <a href="../eval/stat_eval.html">
+
+
+ <b>4.1.</b>
+
+ Statistical evaluation of a prediction model
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="4.1.1" data-path="../eval/auc.html">
+
+ <a href="../eval/auc.html">
+
+
+ <b>4.1.1.</b>
+
+ Area Under the ROC Curve
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="4.2" data-path="../eval/rank.html">
+
+ <a href="../eval/rank.html">
+
+
+ <b>4.2.</b>
+
+ Ranking Measures
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="4.3" data-path="../eval/datagen.html">
+
+ <a href="../eval/datagen.html">
+
+
+ <b>4.3.</b>
+
+ Data Generation
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="4.3.1" data-path="../eval/lr_datagen.html">
+
+ <a href="../eval/lr_datagen.html">
+
+
+ <b>4.3.1.</b>
+
+ Logistic Regression data generation
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part V - Prediction</li>
+
+
+
+ <li class="chapter active" data-level="5.1" data-path="prediction.html">
+
+ <a href="prediction.html">
+
+
+ <b>5.1.</b>
+
+ How Prediction Works
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="5.2" data-path="../regression/general.html">
+
+ <a href="../regression/general.html">
+
+
+ <b>5.2.</b>
+
+ Regression
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="5.3" data-path="../binaryclass/general.html">
+
+ <a href="../binaryclass/general.html">
+
+
+ <b>5.3.</b>
+
+ Binary Classification
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part VI - Binary classification tutorials</li>
+
+
+
+ <li class="chapter " data-level="6.1" data-path="../binaryclass/a9a.html">
+
+ <a href="../binaryclass/a9a.html">
+
+
+ <b>6.1.</b>
+
+ a9a
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.1.1" data-path="../binaryclass/a9a_dataset.html">
+
+ <a href="../binaryclass/a9a_dataset.html">
+
+
+ <b>6.1.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.1.2" data-path="../binaryclass/a9a_lr.html">
+
+ <a href="../binaryclass/a9a_lr.html">
+
+
+ <b>6.1.2.</b>
+
+ Logistic Regression
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.1.3" data-path="../binaryclass/a9a_minibatch.html">
+
+ <a href="../binaryclass/a9a_minibatch.html">
+
+
+ <b>6.1.3.</b>
+
+ Mini-batch Gradient Descent
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.2" data-path="../binaryclass/news20.html">
+
+ <a href="../binaryclass/news20.html">
+
+
+ <b>6.2.</b>
+
+ News20
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.2.1" data-path="../binaryclass/news20_dataset.html">
+
+ <a href="../binaryclass/news20_dataset.html">
+
+
+ <b>6.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2.2" data-path="../binaryclass/news20_pa.html">
+
+ <a href="../binaryclass/news20_pa.html">
+
+
+ <b>6.2.2.</b>
+
+ Perceptron, Passive Aggressive
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2.3" data-path="../binaryclass/news20_scw.html">
+
+ <a href="../binaryclass/news20_scw.html">
+
+
+ <b>6.2.3.</b>
+
+ CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.2.4" data-path="../binaryclass/news20_adagrad.html">
+
+ <a href="../binaryclass/news20_adagrad.html">
+
+
+ <b>6.2.4.</b>
+
+ AdaGradRDA, AdaGrad, AdaDelta
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.3" data-path="../binaryclass/kdd2010a.html">
+
+ <a href="../binaryclass/kdd2010a.html">
+
+
+ <b>6.3.</b>
+
+ KDD2010a
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.3.1" data-path="../binaryclass/kdd2010a_dataset.html">
+
+ <a href="../binaryclass/kdd2010a_dataset.html">
+
+
+ <b>6.3.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.3.2" data-path="../binaryclass/kdd2010a_scw.html">
+
+ <a href="../binaryclass/kdd2010a_scw.html">
+
+
+ <b>6.3.2.</b>
+
+ PA, CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.4" data-path="../binaryclass/kdd2010b.html">
+
+ <a href="../binaryclass/kdd2010b.html">
+
+
+ <b>6.4.</b>
+
+ KDD2010b
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.4.1" data-path="../binaryclass/kdd2010b_dataset.html">
+
+ <a href="../binaryclass/kdd2010b_dataset.html">
+
+
+ <b>6.4.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.4.2" data-path="../binaryclass/kdd2010b_arow.html">
+
+ <a href="../binaryclass/kdd2010b_arow.html">
+
+
+ <b>6.4.2.</b>
+
+ AROW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.5" data-path="../binaryclass/webspam.html">
+
+ <a href="../binaryclass/webspam.html">
+
+
+ <b>6.5.</b>
+
+ Webspam
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="6.5.1" data-path="../binaryclass/webspam_dataset.html">
+
+ <a href="../binaryclass/webspam_dataset.html">
+
+
+ <b>6.5.1.</b>
+
+ Data pareparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="6.5.2" data-path="../binaryclass/webspam_scw.html">
+
+ <a href="../binaryclass/webspam_scw.html">
+
+
+ <b>6.5.2.</b>
+
+ PA1, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="6.6" data-path="../binaryclass/titanic_rf.html">
+
+ <a href="../binaryclass/titanic_rf.html">
+
+
+ <b>6.6.</b>
+
+ Kaggle Titanic
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part VII - Multiclass classification tutorials</li>
+
+
+
+ <li class="chapter " data-level="7.1" data-path="../multiclass/news20.html">
+
+ <a href="../multiclass/news20.html">
+
+
+ <b>7.1.</b>
+
+ News20 Multiclass
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="7.1.1" data-path="../multiclass/news20_dataset.html">
+
+ <a href="../multiclass/news20_dataset.html">
+
+
+ <b>7.1.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.2" data-path="../multiclass/news20_one-vs-the-rest_dataset.html">
+
+ <a href="../multiclass/news20_one-vs-the-rest_dataset.html">
+
+
+ <b>7.1.2.</b>
+
+ Data preparation for one-vs-the-rest classifiers
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.3" data-path="../multiclass/news20_pa.html">
+
+ <a href="../multiclass/news20_pa.html">
+
+
+ <b>7.1.3.</b>
+
+ PA
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.4" data-path="../multiclass/news20_scw.html">
+
+ <a href="../multiclass/news20_scw.html">
+
+
+ <b>7.1.4.</b>
+
+ CW, AROW, SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.5" data-path="../multiclass/news20_ensemble.html">
+
+ <a href="../multiclass/news20_ensemble.html">
+
+
+ <b>7.1.5.</b>
+
+ Ensemble learning
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.1.6" data-path="../multiclass/news20_one-vs-the-rest.html">
+
+ <a href="../multiclass/news20_one-vs-the-rest.html">
+
+
+ <b>7.1.6.</b>
+
+ one-vs-the-rest classifier
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="7.2" data-path="../multiclass/iris.html">
+
+ <a href="../multiclass/iris.html">
+
+
+ <b>7.2.</b>
+
+ Iris
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="7.2.1" data-path="../multiclass/iris_dataset.html">
+
+ <a href="../multiclass/iris_dataset.html">
+
+
+ <b>7.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.2.2" data-path="../multiclass/iris_scw.html">
+
+ <a href="../multiclass/iris_scw.html">
+
+
+ <b>7.2.2.</b>
+
+ SCW
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="7.2.3" data-path="../multiclass/iris_randomforest.html">
+
+ <a href="../multiclass/iris_randomforest.html">
+
+
+ <b>7.2.3.</b>
+
+ RandomForest
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part VIII - Regression tutorials</li>
+
+
+
+ <li class="chapter " data-level="8.1" data-path="../regression/e2006.html">
+
+ <a href="../regression/e2006.html">
+
+
+ <b>8.1.</b>
+
+ E2006-tfidf regression
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="8.1.1" data-path="../regression/e2006_dataset.html">
+
+ <a href="../regression/e2006_dataset.html">
+
+
+ <b>8.1.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.1.2" data-path="../regression/e2006_arow.html">
+
+ <a href="../regression/e2006_arow.html">
+
+
+ <b>8.1.2.</b>
+
+ Passive Aggressive, AROW
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="8.2" data-path="../regression/kddcup12tr2.html">
+
+ <a href="../regression/kddcup12tr2.html">
+
+
+ <b>8.2.</b>
+
+ KDDCup 2012 track 2 CTR prediction
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="8.2.1" data-path="../regression/kddcup12tr2_dataset.html">
+
+ <a href="../regression/kddcup12tr2_dataset.html">
+
+
+ <b>8.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.2.2" data-path="../regression/kddcup12tr2_lr.html">
+
+ <a href="../regression/kddcup12tr2_lr.html">
+
+
+ <b>8.2.2.</b>
+
+ Logistic Regression, Passive Aggressive
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.2.3" data-path="../regression/kddcup12tr2_lr_amplify.html">
+
+ <a href="../regression/kddcup12tr2_lr_amplify.html">
+
+
+ <b>8.2.3.</b>
+
+ Logistic Regression with Amplifier
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="8.2.4" data-path="../regression/kddcup12tr2_adagrad.html">
+
+ <a href="../regression/kddcup12tr2_adagrad.html">
+
+
+ <b>8.2.4.</b>
+
+ AdaGrad, AdaDelta
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part IX - Recommendation</li>
+
+
+
+ <li class="chapter " data-level="9.1" data-path="../recommend/cf.html">
+
+ <a href="../recommend/cf.html">
+
+
+ <b>9.1.</b>
+
+ Collaborative Filtering
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.1.1" data-path="../recommend/item_based_cf.html">
+
+ <a href="../recommend/item_based_cf.html">
+
+
+ <b>9.1.1.</b>
+
+ Item-based Collaborative Filtering
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="9.2" data-path="../recommend/news20.html">
+
+ <a href="../recommend/news20.html">
+
+
+ <b>9.2.</b>
+
+ News20 related article recommendation Tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.2.1" data-path="../multiclass/news20_dataset.html">
+
+ <a href="../multiclass/news20_dataset.html">
+
+
+ <b>9.2.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.2" data-path="../recommend/news20_jaccard.html">
+
+ <a href="../recommend/news20_jaccard.html">
+
+
+ <b>9.2.2.</b>
+
+ LSH/Minhash and Jaccard Similarity
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.3" data-path="../recommend/news20_knn.html">
+
+ <a href="../recommend/news20_knn.html">
+
+
+ <b>9.2.3.</b>
+
+ LSH/Minhash and Brute-Force Search
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.2.4" data-path="../recommend/news20_bbit_minhash.html">
+
+ <a href="../recommend/news20_bbit_minhash.html">
+
+
+ <b>9.2.4.</b>
+
+ kNN search using b-Bits Minhash
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="9.3" data-path="../recommend/movielens.html">
+
+ <a href="../recommend/movielens.html">
+
+
+ <b>9.3.</b>
+
+ MovieLens movie recommendation Tutorial
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="9.3.1" data-path="../recommend/movielens_dataset.html">
+
+ <a href="../recommend/movielens_dataset.html">
+
+
+ <b>9.3.1.</b>
+
+ Data preparation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.2" data-path="../recommend/movielens_cf.html">
+
+ <a href="../recommend/movielens_cf.html">
+
+
+ <b>9.3.2.</b>
+
+ Item-based Collaborative Filtering
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.3" data-path="../recommend/movielens_mf.html">
+
+ <a href="../recommend/movielens_mf.html">
+
+
+ <b>9.3.3.</b>
+
+ Matrix Factorization
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.4" data-path="../recommend/movielens_fm.html">
+
+ <a href="../recommend/movielens_fm.html">
+
+
+ <b>9.3.4.</b>
+
+ Factorization Machine
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="9.3.5" data-path="../recommend/movielens_cv.html">
+
+ <a href="../recommend/movielens_cv.html">
+
+
+ <b>9.3.5.</b>
+
+ 10-fold Cross Validation (Matrix Factorization)
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part X - Anomaly Detection</li>
+
+
+
+ <li class="chapter " data-level="10.1" data-path="../anomaly/lof.html">
+
+ <a href="../anomaly/lof.html">
+
+
+ <b>10.1.</b>
+
+ Outlier Detection using Local Outlier Factor (LOF)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="10.2" data-path="../anomaly/sst.html">
+
+ <a href="../anomaly/sst.html">
+
+
+ <b>10.2.</b>
+
+ Change-Point Detection using Singular Spectrum Transformation (SST)
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="10.3" data-path="../anomaly/changefinder.html">
+
+ <a href="../anomaly/changefinder.html">
+
+
+ <b>10.3.</b>
+
+ ChangeFinder: Detecting Outlier and Change-Point Simultaneously
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XI - Clustering</li>
+
+
+
+ <li class="chapter " data-level="11.1" data-path="../clustering/lda.html">
+
+ <a href="../clustering/lda.html">
+
+
+ <b>11.1.</b>
+
+ Latent Dirichlet Allocation
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="11.2" data-path="../clustering/plsa.html">
+
+ <a href="../clustering/plsa.html">
+
+
+ <b>11.2.</b>
+
+ Probabilistic Latent Semantic Analysis
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XII - GeoSpatial functions</li>
+
+
+
+ <li class="chapter " data-level="12.1" data-path="../geospatial/latlon.html">
+
+ <a href="../geospatial/latlon.html">
+
+
+ <b>12.1.</b>
+
+ Lat/Lon functions
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XIII - Hivemall on Spark</li>
+
+
+
+ <li class="chapter " data-level="13.1" data-path="../spark/getting_started/">
+
+ <a href="../spark/getting_started/">
+
+
+ <b>13.1.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.1.1" data-path="../spark/getting_started/installation.html">
+
+ <a href="../spark/getting_started/installation.html">
+
+
+ <b>13.1.1.</b>
+
+ Installation
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.2" data-path="../spark/binaryclass/">
+
+ <a href="../spark/binaryclass/">
+
+
+ <b>13.2.</b>
+
+ Binary Classification
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.2.1" data-path="../spark/binaryclass/a9a_df.html">
+
+ <a href="../spark/binaryclass/a9a_df.html">
+
+
+ <b>13.2.1.</b>
+
+ a9a Tutorial for DataFrame
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.3" data-path="../spark/binaryclass/">
+
+ <a href="../spark/binaryclass/">
+
+
+ <b>13.3.</b>
+
+ Regression
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.3.1" data-path="../spark/regression/e2006_df.html">
+
+ <a href="../spark/regression/e2006_df.html">
+
+
+ <b>13.3.1.</b>
+
+ E2006-tfidf regression Tutorial for DataFrame
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+ <li class="chapter " data-level="13.4" data-path="../spark/misc/misc.html">
+
+ <a href="../spark/misc/misc.html">
+
+
+ <b>13.4.</b>
+
+ Generic features
+
+ </a>
+
+
+
+ <ul class="articles">
+
+
+ <li class="chapter " data-level="13.4.1" data-path="../spark/misc/topk_join.html">
+
+ <a href="../spark/misc/topk_join.html">
+
+
+ <b>13.4.1.</b>
+
+ Top-k Join processing
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="13.4.2" data-path="../spark/misc/functions.html">
+
+ <a href="../spark/misc/functions.html">
+
+
+ <b>13.4.2.</b>
+
+ Other utility functions
+
+ </a>
+
+
+
+ </li>
+
+
+ </ul>
+
+ </li>
+
+
+
+
+ <li class="header">Part XIV - Hivemall on Docker</li>
+
+
+
+ <li class="chapter " data-level="14.1" data-path="../docker/getting_started.html">
+
+ <a href="../docker/getting_started.html">
+
+
+ <b>14.1.</b>
+
+ Getting Started
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="header">Part XIV - External References</li>
+
+
+
+ <li class="chapter " data-level="15.1" >
+
+ <a target="_blank" href="https://github.com/maropu/hivemall-spark">
+
+
+ <b>15.1.</b>
+
+ Hivemall on Apache Spark
+
+ </a>
+
+
+
+ </li>
+
+ <li class="chapter " data-level="15.2" >
+
+ <a target="_blank" href="https://github.com/daijyc/hivemall/wiki/PigHome">
+
+
+ <b>15.2.</b>
+
+ Hivemall on Apache Pig
+
+ </a>
+
+
+
+ </li>
+
+
+
+
+ <li class="divider"></li>
+
+ <li>
+ <a href="https://www.gitbook.com" target="blank" class="gitbook-link">
+ Published with GitBook
+ </a>
+ </li>
+</ul>
+
+
+ </nav>
+
+
+ </div>
+
+ <div class="book-body">
+
+ <div class="body-inner">
+
+
+
+<div class="book-header" role="navigation">
+
+
+ <!-- Title -->
+ <h1>
+ <i class="fa fa-circle-o-notch fa-spin"></i>
+ <a href=".." >How Prediction Works</a>
+ </h1>
+</div>
+
+
+
+
+ <div class="page-wrapper" tabindex="-1" role="main">
+ <div class="page-inner">
+
+<div id="book-search-results">
+ <div class="search-noresults">
+
+ <section class="normal markdown-section">
+
+ <!--
+ Licensed to the Apache Software Foundation (ASF) under one
+ or more contributor license agreements. See the NOTICE file
+ distributed with this work for additional information
+ regarding copyright ownership. The ASF licenses this file
+ to you under the Apache License, Version 2.0 (the
+ "License"); you may not use this file except in compliance
+ with the License. You may obtain a copy of the License at
+
+ http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing,
+ software distributed under the License is distributed on an
+ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ KIND, either express or implied. See the License for the
+ specific language governing permissions and limitations
+ under the License.
+-->
+<!-- toc --><div id="toc" class="toc">
+
+<ul>
+<li><a href="#what-is-prediction-problem">What is "prediction problem"?</a></li>
+<li><a href="#regression">Regression</a></li>
+<li><a href="#classification">Classification</a></li>
+<li><a href="#mathematical-formulation-of-generic-prediction-model">Mathematical formulation of generic prediction model</a></li>
+</ul>
+
+</div><!-- tocstop -->
+<h1 id="what-is-prediction-problem">What is "prediction problem"?</h1>
+<p>In a context of machine learning, numerous tasks can be seen as <strong>prediction problem</strong>. For example, this user guide provides solutions for:</p>
+<ul>
+<li><a href="../binaryclass/webspam.html">spam detection</a></li>
+<li><a href="../multiclass/news20.html">news article classification</a></li>
+<li><a href="../regression/kddcup12tr2.html">click-through-rate estimation</a></li>
+</ul>
+<p>For any kinds of prediction problems, we generally provide a set of input-output pairs as:</p>
+<ul>
+<li><strong>Input:</strong> Set of features<ul>
+<li>e.g., <code>["1:0.001","4:0.23","35:0.0035",...]</code></li>
+</ul>
+</li>
+<li><strong>Output:</strong> Target value<ul>
+<li>e.g., 1, 0, 0.54, 42.195, ...</li>
+</ul>
+</li>
+</ul>
+<p>Once a prediction model has been constructed based on the samples, the model can make prediction for unforeseen inputs. </p>
+<p>In order to train prediction models, an algorithm so-called <strong><em>stochastic gradient descent</em></strong> (SGD) is normally applied. You can learn more about this from the following external resources:</p>
+<ul>
+<li><a href="http://scikit-learn.org/stable/modules/sgd.html" target="_blank">scikit-learn documentation</a></li>
+<li><a href="http://spark.apache.org/docs/latest/mllib-optimization.html" target="_blank">Spark MLlib documentation</a></li>
+</ul>
+<p>Importantly, depending on types of output value, prediction problem can be categorized into <strong>regression</strong> and <strong>classification</strong> problem.</p>
+<h1 id="regression">Regression</h1>
+<p>The goal of regression is to predict <strong>real values</strong> as shown below:</p>
+<table>
+<thead>
+<tr>
+<th style="text-align:left">features (input)</th>
+<th style="text-align:center">target real value (output)</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td style="text-align:left">["1:0.001","4:0.23","35:0.0035",...]</td>
+<td style="text-align:center">21.3</td>
+</tr>
+<tr>
+<td style="text-align:left">["1:0.2","3:0.1","13:0.005",...]</td>
+<td style="text-align:center">6.2</td>
+</tr>
+<tr>
+<td style="text-align:left">["5:1.3","22:0.0.089","77:0.0001",...]</td>
+<td style="text-align:center">17.1</td>
+</tr>
+<tr>
+<td style="text-align:left">...</td>
+<td style="text-align:center">...</td>
+</tr>
+</tbody>
+</table>
+<p>In practice, target values could be any of small/large float/int negative/positive values. <a href="../regression/kddcup12tr2.html">Our CTR prediction tutorial</a> solves regression problem with small floating point target values in a 0-1 range, for example.</p>
+<p>While there are several ways to realize regression by using Hivemall, <code>train_regression()</code> is one of the most flexible functions. This feature is explained in: <a href="../regression/general.html">Regression</a>.</p>
+<h1 id="classification">Classification</h1>
+<p>In contrast to regression, output for classification problems should be (integer) <strong>labels</strong>:</p>
+<table>
+<thead>
+<tr>
+<th style="text-align:left">features (input)</th>
+<th style="text-align:center">label (output)</th>
+</tr>
+</thead>
+<tbody>
+<tr>
+<td style="text-align:left">["1:0.001","4:0.23","35:0.0035",...]</td>
+<td style="text-align:center">0</td>
+</tr>
+<tr>
+<td style="text-align:left">["1:0.2","3:0.1","13:0.005",...]</td>
+<td style="text-align:center">1</td>
+</tr>
+<tr>
+<td style="text-align:left">["5:1.3","22:0.0.089","77:0.0001",...]</td>
+<td style="text-align:center">1</td>
+</tr>
+<tr>
+<td style="text-align:left">...</td>
+<td style="text-align:center">...</td>
+</tr>
+</tbody>
+</table>
+<p>In case the number of possible labels is 2 (0/1 or -1/1), the problem is <strong>binary classification</strong>, and Hivemall's <code>train_classifier()</code> function enables you to build binary classifiers. <a href="../binaryclass/general.html">Binary Classification</a> demonstrates how to use the function.</p>
+<p>Another type of classification problems is <strong>multi-class classification</strong>. This task assumes that the number of possible labels is more than 2. We need to use different functions for the multi-class problems, and our <a href="../multiclass/news20.html">news20</a> and <a href="../multiclass/iris.html">iris</a> tutorials would be helpful.</p>
+<h1 id="mathematical-formulation-of-generic-prediction-model">Mathematical formulation of generic prediction model</h1>
+<p>Here, we briefly explain about how prediction model is constructed.</p>
+<p>First and foremost, we represent <strong>input</strong> and <strong>output</strong> for prediction models as follows:</p>
+<ul>
+<li><strong>Input:</strong> a vector <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">x</mi></mrow></mrow><annotation encoding="application/x-tex">\mathbf{x}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span></span></span></span></li>
+<li><strong>Output:</strong> a value <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>y</mi></mrow><annotation encoding="application/x-tex">y</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.43056em;"></span><span class="strut bottom" style="height:0.625em;vertical-align:-0.19444em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.03588em;">y</span></span></span></span></li>
+</ul>
+<p>For a set of samples <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mo>(</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mn>1</mn></msub><mo separator="true">,</mo><msub><mi>y</mi><mn>1</mn></msub><mo>)</mo><mo separator="true">,</mo><mo>(</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mn>2</mn></msub><mo separator="true">,</mo><msub><mi>y</mi><mn>2</mn></msub><mo>)</mo><mo separator="true">,</mo><mo>⋯</mo><mo separator="true">,</mo><mo>(</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mi>n</mi></msub><mo separator="true">,</mo><msub><mi>y</mi><mi>n</mi></msub><mo>)</mo></mrow><annotation encoding="application/x-tex">(\mathbf{x}_1, y_1), (\mathbf{x}_2, y_2), \cdots, (\mathbf{x}_n, y_n)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.75em;"></span><span class="strut bottom" style="height:1em;vertical-align:-0.25em;"></span><span class="base textstyle uncramped"><spa
n class="mopen">(</span><span class="mord"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtight">1</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtig
ht">1</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mclose">)</span><span class="mpunct">,</span><span class="mopen">(</span><span class="mord"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtight">2</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><spa
n style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathrm mtight">2</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mclose">)</span><span class="mpunct">,</span><span class="minner">⋯</span><span class="mpunct">,</span><span class="mopen">(</span><span class="mord"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">n</span></span></span><span c
lass="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">n</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mclose">)</span></span></span></span>, the goal of prediction algorithms is to find a weight vector (i.e., parameters) <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">w</mi></mr
ow></mrow><annotation encoding="application/x-tex">\mathbf{w}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span></span></span></span> by minimizing the following error:</p>
+<p><span class="katex-display"><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>E</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo>)</mo><mo>:</mo><mo>=</mo><mfrac><mrow><mn>1</mn></mrow><mrow><mi>n</mi></mrow></mfrac><msubsup><mo>∑</mo><mrow><mi>i</mi><mo>=</mo><mn>1</mn></mrow><mrow><mi>n</mi></mrow></msubsup><mi>L</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo separator="true">;</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mi>i</mi></msub><mo separator="true">,</mo><msub><mi>y</mi><mi>i</mi></msub><mo>)</mo><mo>+</mo><mi>λ</mi><mi>R</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo>)</mo></mrow><annotation encoding="application/x-tex">
+E(\mathbf{w}) := \frac{1}{n} \sum_{i=1}^{n} L(\mathbf{w}; \mathbf{x}_i, y_i) + \lambda R(\mathbf{w})
+</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:1.6513970000000002em;"></span><span class="strut bottom" style="height:2.929066em;vertical-align:-1.277669em;"></span><span class="base displaystyle textstyle uncramped"><span class="mord mathit" style="margin-right:0.05764em;">E</span><span class="mopen">(</span><span class="mord displaystyle textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mclose">)</span><span class="mrel">:</span><span class="mrel">=</span><span class="mord reset-textstyle displaystyle textstyle uncramped"><span class="mopen sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span><span class="mfrac"><span class="vlist"><span style="top:0.686em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle cramped"><span class="mord texts
tyle cramped"><span class="mord mathit">n</span></span></span></span><span style="top:-0.22999999999999998em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped frac-line"></span></span><span style="top:-0.677em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathrm">1</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="mclose sizing reset-size5 size5 reset-textstyle textstyle uncramped nulldelimiter"></span></span><span class="mop op-limits"><span class="vlist"><span style="top:1.1776689999999999em;margin-left:0em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-siz
e:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord scriptstyle cramped mtight"><span class="mord mathit mtight">i</span><span class="mrel mtight">=</span><span class="mord mathrm mtight">1</span></span></span></span><span style="top:-0.000005000000000143778em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span><span class="mop op-symbol large-op">∑</span></span></span><span style="top:-1.2500050000000003em;margin-left:0em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle uncramped mtight"><span class="mord scriptstyle uncramped mtight"><span class="mord mathit mtight">n</span></span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span><span class="mord mathit">L
</span><span class="mopen">(</span><span class="mord displaystyle textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mpunct">;</span><span class="mord"><span class="mord displaystyle textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span cl
ass="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mclose">)</span><span class="mbin">+</span><span class="mord mathit">λ</span><span class="mord mathit" style="margin-right:0.00773em;">R</span><span class="mopen">(</span><span class="mord displaystyle textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mclose">)</span></span></span></span></span></p>
+<p>In the above formulation, there are two auxiliary functions we have to know: </p>
+<ul>
+<li><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>L</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo separator="true">;</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mi>i</mi></msub><mo separator="true">,</mo><msub><mi>y</mi><mi>i</mi></msub><mo>)</mo></mrow><annotation encoding="application/x-tex">L(\mathbf{w}; \mathbf{x}_i, y_i)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.75em;"></span><span class="strut bottom" style="height:1em;vertical-align:-0.25em;"></span><span class="base textstyle uncramped"><span class="mord mathit">L</span><span class="mopen">(</span><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mpunct">;</span><span class="mord"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;ma
rgin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><s
pan class="mclose">)</span></span></span></span><ul>
+<li><strong>Loss function</strong> for a single sample <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mo>(</mo><msub><mrow><mi mathvariant="bold">x</mi></mrow><mi>i</mi></msub><mo separator="true">,</mo><msub><mi>y</mi><mi>i</mi></msub><mo>)</mo></mrow><annotation encoding="application/x-tex">(\mathbf{x}_i, y_i)</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.75em;"></span><span class="strut bottom" style="height:1em;vertical-align:-0.25em;"></span><span class="base textstyle uncramped"><span class="mopen">(</span><span class="mord"><span class="mord textstyle uncramped"><span class="mord mathbf">x</span></span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</spa
n></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mpunct">,</span><span class="mord"><span class="mord mathit" style="margin-right:0.03588em;">y</span><span class="msupsub"><span class="vlist"><span style="top:0.15em;margin-right:0.05em;margin-left:-0.03588em;"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span><span class="reset-textstyle scriptstyle cramped mtight"><span class="mord mathit mtight">i</span></span></span><span class="baseline-fix"><span class="fontsize-ensurer reset-size5 size5"><span style="font-size:0em;">​</span></span>​</span></span></span></span><span class="mclose">)</span></span></span></span> and given <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">w</mi></mrow></mrow><annotation encoding="application/x-te
x">\mathbf{w}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span></span></span></span>.</li>
+<li>If this function produces small values, it means the parameter <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">w</mi></mrow></mrow><annotation encoding="application/x-tex">\mathbf{w}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span></span></span></span> is successfully learnt. </li>
+</ul>
+</li>
+<li><span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>R</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo>)</mo></mrow><annotation encoding="application/x-tex">R(\mathbf{w})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.75em;"></span><span class="strut bottom" style="height:1em;vertical-align:-0.25em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.00773em;">R</span><span class="mopen">(</span><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mclose">)</span></span></span></span><ul>
+<li><strong>Regularization function</strong> for the current parameter <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">w</mi></mrow></mrow><annotation encoding="application/x-tex">\mathbf{w}</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span></span></span></span>.</li>
+<li>It prevents failing to a negative condition so-called <strong>over-fitting</strong>.</li>
+</ul>
+</li>
+</ul>
+<p>(<span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>λ</mi></mrow><annotation encoding="application/x-tex">\lambda</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.69444em;"></span><span class="strut bottom" style="height:0.69444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord mathit">λ</span></span></span></span> is a small value which controls the effect of regularization function.)</p>
+<p>Eventually, minimizing the function <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mi>E</mi><mo>(</mo><mrow><mi mathvariant="bold">w</mi></mrow><mo>)</mo></mrow><annotation encoding="application/x-tex">E(\mathbf{w})</annotation></semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.75em;"></span><span class="strut bottom" style="height:1em;vertical-align:-0.25em;"></span><span class="base textstyle uncramped"><span class="mord mathit" style="margin-right:0.05764em;">E</span><span class="mopen">(</span><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span><span class="mclose">)</span></span></span></span> can be implemented by the SGD technique as described before, and <span class="katex"><span class="katex-mathml"><math><semantics><mrow><mrow><mi mathvariant="bold">w</mi></mrow></mrow><annotation encoding="application/x-tex">\mathbf{w}</annotation><
/semantics></math></span><span class="katex-html" aria-hidden="true"><span class="strut" style="height:0.44444em;"></span><span class="strut bottom" style="height:0.44444em;vertical-align:0em;"></span><span class="base textstyle uncramped"><span class="mord textstyle uncramped"><span class="mord mathbf" style="margin-right:0.01597em;">w</span></span></span></span></span> itself is used as a "model" for future prediction.</p>
+<p>Interestingly, depending on a choice of loss and regularization function, prediction model you obtained will behave differently; even if one combination could work as a classifier, another choice might be appropriate for regression.</p>
+<p>Below we list possible options for <code>train_regression</code> and <code>train_classifier</code>, and this is the reason why these two functions are the most flexible in Hivemall:</p>
+<ul>
+<li>Loss function: <code>-loss</code>, <code>-loss_function</code><ul>
+<li>For <code>train_regression</code><ul>
+<li>SquaredLoss</li>
+<li>QuantileLoss</li>
+<li>EpsilonInsensitiveLoss</li>
+<li>SquaredEpsilonInsensitiveLoss</li>
+<li>HuberLoss</li>
+</ul>
+</li>
+<li>For <code>train_classifier</code><ul>
+<li>HingeLoss</li>
+<li>LogLoss</li>
+<li>SquaredHingeLoss</li>
+<li>ModifiedHuberLoss</li>
+<li>SquaredLoss</li>
+<li>QuantileLoss</li>
+<li>EpsilonInsensitiveLoss</li>
+<li>SquaredEpsilonInsensitiveLoss</li>
+<li>HuberLoss</li>
+</ul>
+</li>
+</ul>
+</li>
+<li>Regularization function: <code>-reg</code>, <code>-regularization</code><ul>
+<li>L1</li>
+<li>L2</li>
+<li>ElasticNet</li>
+<li>RDA</li>
+</ul>
+</li>
+</ul>
+<p>Additionally, there are several variants of the SGD technique, and it is also configureable as:</p>
+<ul>
+<li>Optimizer <code>-opt</code>, <code>-optimizer</code><ul>
+<li>SGD</li>
+<li>AdaGrad</li>
+<li>AdaDelta</li>
+<li>Adam</li>
+</ul>
+</li>
+</ul>
+<p>In practice, you can try different combinations of the options in order to achieve higher prediction accuracy.
+<div id="page-footer" class="localized-footer"><hr><!--
+ Licensed to the Apache Software Foundation (ASF) under one
+ or more contributor license agreements. See the NOTICE file
+ distributed with this work for additional information
+ regarding copyright ownership. The ASF licenses this file
+ to you under the Apache License, Version 2.0 (the
+ "License"); you may not use this file except in compliance
+ with the License. You may obtain a copy of the License at
+
+ http://www.apache.org/licenses/LICENSE-2.0
+
+ Unless required by applicable law or agreed to in writing,
+ software distributed under the License is distributed on an
+ "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ KIND, either express or implied. See the License for the
+ specific language governing permissions and limitations
+ under the License.
+-->
+<p><sub><font color="gray">
+Apache Hivemall is an effort undergoing incubation at The Apache Software Foundation (ASF), sponsored by the Apache Incubator.
+</font></sub></p>
+</div></p>
+
+
+ </section>
+
+ </div>
+ <div class="search-results">
+ <div class="has-results">
+
+ <h1 class="search-results-title"><span class='search-results-count'></span> results matching "<span class='search-query'></span>"</h1>
+ <ul class="search-results-list"></ul>
+
+ </div>
+ <div class="no-results">
+
+ <h1 class="search-results-title">No results matching "<span class='search-query'></span>"</h1>
+
+ </div>
+ </div>
+</div>
+
+ </div>
+ </div>
+
+ </div>
+
+
+
+
+ </div>
+
+ <script>
+ var gitbook = gitbook || [];
+ gitbook.push(function() {
+ gitbook.page.hasChanged({"page":{"title":"How Prediction Works","level":"5.1","depth":1,"next":{"title":"Regression","level":"5.2","depth":1,"path":"regression/general.md","ref":"regression/general.md","articles":[]},"previous":{"title":"Logistic Regression data generation","level":"4.3.1","depth":2,"path":"eval/lr_datagen.md","ref":"eval/lr_datagen.md","articles":[]},"dir":"ltr"},"config":{"plugins":["theme-api","edit-link","github","splitter","sitemap","etoc","callouts","toggle-chapters","anchorjs","codeblock-filename","expandable-chapters","multipart","codeblock-filename","katex","emphasize","localized-footer"],"styles":{"website":"styles/website.css","pdf":"styles/pdf.css","epub":"styles/epub.css","mobi":"styles/mobi.css","ebook":"styles/ebook.css","print":"styles/print.css"},"pluginsConfig":{"emphasize":{},"callouts":{},"etoc":{"h2lb":3,"header":1,"maxdepth":3,"mindepth":1,"notoc":true},"github":{"url":"https://github.com/apache/incubator-hivemall/"},"splitter":{},"
search":{},"downloadpdf":{"base":"https://github.com/apache/incubator-hivemall/docs/gitbook","label":"PDF","multilingual":false},"multipart":{},"localized-footer":{"filename":"FOOTER.md","hline":"true"},"lunr":{"maxIndexSize":1000000,"ignoreSpecialCharacters":false},"katex":{},"fontsettings":{"theme":"white","family":"sans","size":2,"font":"sans"},"highlight":{},"codeblock-filename":{},"sitemap":{"hostname":"http://hivemall.incubator.apache.org/"},"theme-api":{"languages":[],"split":false,"theme":"dark"},"sharing":{"facebook":true,"twitter":true,"google":false,"weibo":false,"instapaper":false,"vk":false,"all":["facebook","google","twitter","weibo","instapaper"]},"edit-link":{"label":"Edit","base":"https://github.com/apache/incubator-hivemall/docs/gitbook"},"theme-default":{"styles":{"website":"styles/website.css","pdf":"styles/pdf.css","epub":"styles/epub.css","mobi":"styles/mobi.css","ebook":"styles/ebook.css","print":"styles/print.css"},"showLevel":true},"anchorjs":{"selector":"h1
,h2,h3,*:not(.callout) > h4,h5"},"toggle-chapters":{},"expandable-chapters":{}},"theme":"default","pdf":{"pageNumbers":true,"fontSize":12,"fontFamily":"Arial","paperSize":"a4","chapterMark":"pagebreak","pageBreaksBefore":"/","margin":{"right":62,"left":62,"top":56,"bottom":56}},"structure":{"langs":"LANGS.md","readme":"README.md","glossary":"GLOSSARY.md","summary":"SUMMARY.md"},"variables":{},"title":"Hivemall User Manual","links":{"sidebar":{"<i class=\"fa fa-home\"></i> Home":"http://hivemall.incubator.apache.org/"}},"gitbook":"3.x.x","description":"User Manual for Apache Hivemall"},"file":{"path":"misc/prediction.md","mtime":"2017-06-15T10:18:54.000Z","type":"markdown"},"gitbook":{"version":"3.2.2","time":"2017-06-15T10:33:21.138Z"},"basePath":"..","book":{"language":""}});
+ });
+ </script>
+</div>
+
+
+ <script src="../gitbook/gitbook.js"></script>
+ <script src="../gitbook/theme.js"></script>
+
+
+ <script src="../gitbook/gitbook-plugin-edit-link/plugin.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-github/plugin.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-splitter/splitter.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-etoc/plugin.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-toggle-chapters/toggle.js"></script>
+
+
+
+ <script src="https://cdnjs.cloudflare.com/ajax/libs/anchor-js/3.1.1/anchor.min.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-anchorjs/anchor-style.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-expandable-chapters/expandable-chapters.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-search/search-engine.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-search/search.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-lunr/lunr.min.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-lunr/search-lunr.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-sharing/buttons.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-fontsettings/fontsettings.js"></script>
+
+
+
+ <script src="../gitbook/gitbook-plugin-theme-api/theme-api.js"></script>
+
+
+
+ </body>
+</html>
+