Is there a way to use a Random Forest model on Android to make predictions?

Question

I've been looking at primarily Weka to do machine learning testing, and I've found that Random Forest models have the best results for my purposes. I wanted to save this model and implement it on Android to predict new values, but these days it seems nigh-impossible to implement Weka on Android (all the available sources are super out-dated).

Are there any ways to train a Random Forest model, and then implement that model on an Android app to predict new values?

user1808924 user1808924 · Accepted Answer · 2018-04-13T06:32:41

In addition to Weka, you can train Random Forest (RF) models also with R, Scikit-Learn or Apache Spark ML. You can export/convert RF models from their native representation into the standardized PMML representation using R2PMML, SkLearn2PMML or JPMML-SparkML-Package tools, respectively, and then import and score such models using Java PMML scoring engines such as JPMML-Evaluator. The latter has a direct Android integration available.

Is there a way to use a Random Forest model on Android to make predictions?

2 Answers