From 9cca3dbf4add9004a769dee1a556987e37230294 Mon Sep 17 00:00:00 2001
From: Nick Pentreath <nickp@za.ibm.com>
Date: Thu, 2 Mar 2017 15:51:16 +0200
Subject: [PATCH] [SPARK-19345][ML][DOC] Add doc for "coldStartStrategy" usage
in ALS
[SPARK-14489](https://issues.apache.org/jira/browse/SPARK-14489) added the ability to skip `NaN` predictions during `ALSModel.transform`. This PR adds documentation for the `coldStartStrategy` param to the ALS user guide, and add code to the examples to illustrate usage.
## How was this patch tested?
Doc and example change only. Build HTML doc locally and verified example code builds, and runs in shell for Scala/Python.
Author: Nick Pentreath <nickp@za.ibm.com>
Closes #17102 from MLnick/SPARK-19345-coldstart-doc.
---
docs/ml-collaborative-filtering.md | 28 +++++++++++++++++++
.../spark/examples/ml/JavaALSExample.java | 2 ++
examples/src/main/python/ml/als_example.py | 4 ++-
.../apache/spark/examples/ml/ALSExample.scala | 2 ++
4 files changed, 35 insertions(+), 1 deletion(-)
diff --git a/docs/ml-collaborative-filtering.md b/docs/ml-collaborative-filtering.md
index cfe835172a..58f2d4b531 100644
--- a/docs/ml-collaborative-filtering.md
+++ b/docs/ml-collaborative-filtering.md
@@ -59,6 +59,34 @@ This approach is named "ALS-WR" and discussed in the paper
It makes `regParam` less dependent on the scale of the dataset, so we can apply the
best parameter learned from a sampled subset to the full dataset and expect similar performance.
+### Cold-start strategy
+
+When making predictions using an `ALSModel`, it is common to encounter users and/or items in the
+test dataset that were not present during training the model. This typically occurs in two
+scenarios:
+
+1. In production, for new users or items that have no rating history and on which the model has not
+been trained (this is the "cold start problem").
+2. During cross-validation, the data is split between training and evaluation sets. When using
+simple random splits as in Spark's `CrossValidator` or `TrainValidationSplit`, it is actually
+very common to encounter users and/or items in the evaluation set that are not in the training set
+
+By default, Spark assigns `NaN` predictions during `ALSModel.transform` when a user and/or item
+factor is not present in the model. This can be useful in a production system, since it indicates
+a new user or item, and so the system can make a decision on some fallback to use as the prediction.
+
+However, this is undesirable during cross-validation, since any `NaN` predicted values will result
+in `NaN` results for the evaluation metric (for example when using `RegressionEvaluator`).
+This makes model selection impossible.
+
+Spark allows users to set the `coldStartStrategy` parameter
+to "drop" in order to drop any rows in the `DataFrame` of predictions that contain `NaN` values.
+The evaluation metric will then be computed over the non-`NaN` data and will be valid.
+Usage of this parameter is illustrated in the example below.
+
+**Note:** currently the supported cold start strategies are "nan" (the default behavior mentioned
+above) and "drop". Further strategies may be supported in future.
+
**Examples**
<div class="codetabs">
diff --git a/examples/src/main/java/org/apache/spark/examples/ml/JavaALSExample.java b/examples/src/main/java/org/apache/spark/examples/ml/JavaALSExample.java
index 33ba668b32..81970b7c81 100644
--- a/examples/src/main/java/org/apache/spark/examples/ml/JavaALSExample.java
+++ b/examples/src/main/java/org/apache/spark/examples/ml/JavaALSExample.java
@@ -103,6 +103,8 @@ public class JavaALSExample {
ALSModel model = als.fit(training);
// Evaluate the model by computing the RMSE on the test data
+ // Note we set cold start strategy to 'drop' to ensure we don't get NaN evaluation metrics
+ model.setColdStartStrategy("drop");
Dataset<Row> predictions = model.transform(test);
RegressionEvaluator evaluator = new RegressionEvaluator()
diff --git a/examples/src/main/python/ml/als_example.py b/examples/src/main/python/ml/als_example.py
index 1a979ff5b5..2e7214ed56 100644
--- a/examples/src/main/python/ml/als_example.py
+++ b/examples/src/main/python/ml/als_example.py
@@ -44,7 +44,9 @@ if __name__ == "__main__":
(training, test) = ratings.randomSplit([0.8, 0.2])
# Build the recommendation model using ALS on the training data
- als = ALS(maxIter=5, regParam=0.01, userCol="userId", itemCol="movieId", ratingCol="rating")
+ # Note we set cold start strategy to 'drop' to ensure we don't get NaN evaluation metrics
+ als = ALS(maxIter=5, regParam=0.01, userCol="userId", itemCol="movieId", ratingCol="rating",
+ coldStartStrategy="drop")
model = als.fit(training)
# Evaluate the model by computing the RMSE on the test data
diff --git a/examples/src/main/scala/org/apache/spark/examples/ml/ALSExample.scala b/examples/src/main/scala/org/apache/spark/examples/ml/ALSExample.scala
index bb5d163608..868f49b16f 100644
--- a/examples/src/main/scala/org/apache/spark/examples/ml/ALSExample.scala
+++ b/examples/src/main/scala/org/apache/spark/examples/ml/ALSExample.scala
@@ -65,6 +65,8 @@ object ALSExample {
val model = als.fit(training)
// Evaluate the model by computing the RMSE on the test data
+ // Note we set cold start strategy to 'drop' to ensure we don't get NaN evaluation metrics
+ model.setColdStartStrategy("drop")
val predictions = model.transform(test)
val evaluator = new RegressionEvaluator()
--
GitLab